Abstract
Forest soil sediment disasters, categorized as landslides, slow-moving landslides (land creep), and debris flows, result in destruction of forests and properties, loss of life, and damage to infrastructures. Therefore, it is imperative to evaluate factors that cause slow-moving landslides and establish a national-level hazard information system. In this study, we used geographic information system (GIS) to investigate the difference between landslides and slow-moving landslides by overlapping and analyzing landslide hazard map with slow-moving landslide sites. Slow-moving landslide areas (SLAs) and non-SLAs were characterized to compare the effects of altitude and slope in slow-moving landslide events. Finally, we performed a hydrographical analysis of waterways to compare the SLAs and non-SLAs. Grade 3 slow-moving landslide sites were found to be the most prevalent among all landslide hazard grades (1–5), including those in regions outside the specified grades. The altitude and slope of SLAs were 5 m higher and 1° steeper than those of non-SLAs, respectively. The waterways in non-SLAs (205.4 m) were longer than those in SLAs (85.2 m). Our study highlights the importance of altitude, slope, and waterways as triggers for slow-moving landslides and provides an empirical basis for establishing national hazard information systems for these events.
1 Introduction
Natural disasters are becoming increasingly prevalent worldwide (). In particular, large-scale forest soil sediment disasters induced by complex topographical and climatic factors destroy forests and cause loss of life, property destruction, and infrastructure damage (; ). According to a World Bank report, approximately 300 million people reside in landslide-prone areas, with approximately 600 annual landslide-related mortalities (). Numerous forest soil sediment disasters occurred in South Korea in 2000 as a result of typhoons and torrential rain (), and related damages were particularly severe in the summers of 2011 and 2020 owing to heavy precipitation ().
Forest soil sediment disasters are categorized as landslides, slow-moving landslides (land creep), and debris flows in South Korea. The United States Geological Survey (USGS) report, sponsored by the International Consortium on Landslides, defines a landslide as the downslope movement of soil, rock, and organic material under the influence of gravity and the resulting deformation of the landscape. The USGS explained that the term landslide is a general term used to describe this type of mass movement. Landslides are classified into various kinds, including falls (rockfalls, topples), slides (rotational landslides, translational landslides), spreads (lateral spreads), and flows (debris flows, lahars (volcanic debris flows), debris avalanches, earthflows, slow earthflows (creep), and flows in permafrost) (). Among these, South Korea’s slow-moving landslide can be explained as slow earthflow (creep). Here, creep is an informal name for slow earthflow and refers to a phenomenon in which the soil or rocks forming a slope move downward so slowly and steadily that they are imperceptible. In other words, movement is caused by internal shear stress, which is enough to deform the terrain but not strong enough to cause rapid damage. Additionally, creep is widespread worldwide and is the most common type of landslide, often preceding faster and more destructive types of landslides (Highland et al., 2008).
The erosion of deep soils of formerly mountainous forest areas and increased urbanization and forest land-use conversion have increased the occurrence of slow-moving landslides recently (). The stability of areas at hazard of slow-moving landslides is indirectly influenced by ground and effective pores, which determine rainfall infiltration and groundwater levels. Groundwater levels and pore water pressure rise with increased precipitation, which reduces the shearing force of the soil (; ; ; ) and can lead to slow-moving landslides.
Several major slow-moving landslides during the early 1990s in South Korea were initially classified as landslides (). Slow-moving landslides and landslides differ in their geological features, soil quality, earth and rock movement velocities, topography, activity and movement status, clod, occasional cause, scale, outcomes, and slope (). Landslides entail an increase in groundwater level and the eventual downward sliding of soil over bedrock (including the sliding surface of the soil); they are characterized by the instantaneous collapse of sediments because of the lubricating action of water at the soil–bedrock interface. In contrast, slow-moving landslides involve the slow movement of the entire soil layer (approximately 0.01 mm/day) owing to the impact of the clay layer in the soil stratum or groundwater pressure on the underlying rock ().
The occurrence of slow-moving landslides is determined by a combination of internal (e.g., geology, soil composition, topography, hydrology, and forest floor structure) and external (e.g., rainfall, earthquakes, and large-scale construction) factors. A national-level slow-moving landslide vulnerability assessment and hazard information system (slow-moving landslide hazard map) must be established to effectively observe, analyze, and predict the hazard of slow-moving landslide occurrence. South Korea currently provides landslide hazard maps to identify vulnerable areas, but there are no instruments to predict the occurrence of slow-moving landslides. Furthermore, recent studies have analyzed the geology (; ; ), topography (; ; ; ), rainfall (; ), groundwater (; ; ), reservoir levels (), forest floor (; ; ), and slope (; ) of slow-moving landslides. However, these studies failed to consider the differences in characteristics depending on slow-moving landslide triggering factors between slow-moving landslide areas (SLAs) and surrounding non-SLAs, which are essential for predicting the overall hazard and outcome of slow-moving landslides. Given their importance, the altitude and slope differences between mountainous SLAs and non-SLAs should be considered when creating slow-moving landslide hazard maps.
In contrast to landslides caused by strong earthquakes or heavy rains, silent large-scale landslides in Zhaobishan, China, are said to occur in large water confluence areas behind the main scarp in mountainous areas (). However, no study has yet been conducted to compare and analyze the differences in waterways within mountainous areas due to rainfall, which can affect changes in groundwater levels, a significant factor in slow-moving landslides. The principle of water flowing from high to low altitudes and from steep to shallow slopes has been used to analyze waterway behavior via a digital elevation model (DEM) and geographic information system (GIS) modeling ().
In this study, we evaluated the applicability of a hazard information system for slow-moving landslides in South Korea. Accordingly, GIS modeling was utilized to overlap and analyze landslide hazard map and slow-moving landslide sites. We were interested in areas where slow-moving landslides had caused casualties or property damage in mountainous areas near our living areas. We also focused on why slow-moving landslides occur on some of these mountain slopes. It is essential to clearly identify the causes of slow-moving landslides, one of the forest soil sediment disasters that is likely to be caused by a complex mechanism due to numerous factors. Prior to this, in this study, we used GIS to attempt to verify whether topographical characteristics of mountainous areas, such as altitude, slope, and the characteristics of waterways formed by rainfall and topographical features, act as characteristic factors causing slow-moving landslides. To this end, we identified SLAs and their adjacent non-SLAs to analyze the altitude and slope differences. We then differentiated SLAs and non-SLAs based on the waterways created by precipitation and torrential rains to determine the applicability of waterways as novel triggering factors for slow-moving landslides.
2 Materials and methods
2.1 Selection of study sites
We selected 12 study sites across South Korea where recent slow-moving landslides caused mortalities or property damage (). However, a digital topographic map was unavailable for two sites prior to the estimated period of slow-moving landslide occurrence; therefore, 10 sites were selected for the final analysis (Figure 1).
FIGURE 1
2.2 Setting analysis area
To analyze the differences in altitude, slope, and waterway between SLAs and non-SLAs, we first established the polygon boundary of the slow-moving landslide site using a Pro-XRS GPS device (Trimble, Westminster, Colorado, United States) with a positioning accuracy of 50 cm or less (post-processed). The SLAs were created using the shape creation function of GIS software to form rectangular polygons around the boundaries of slow-moving landslide sites. The adjacent non-SLA was positioned on the left and right sides of half the corresponding SLA based on the direction and length of the SLA; placement was determined by the lowest end contour line of the SLA (See Figure 2). To understand the characteristics of SLAs, we established SLAs surrounding the boundaries of slow-moving landslide sites and non-SLAs adjacent to the SLAs. In addition, the size and placement of SLAs and non-SLAs were determined to ensure fairness in statistical analysis by ensuring equality of samples between comparison groups. Meanwhile, ArcMap 10.8 (ESRI, Redlands, CA, United States) was used to set up the analysis area, conduct overlap analyses, and analyze altitude, slope, and waterway.
FIGURE 2
2.3 Overlap analysis of the landslide hazard map and slow-moving landslide site
We overlapped the landslide hazard map (1:25,000; provided byKorea Forest Service; compiled in 2019) and the slow-moving landslide site boundary (n = 10) to analyze the differences. The landslide hazard map was clipped using the slow-moving landslide site boundary for the overlap analysis, and the area by each grade was calculated. The landslide hazard map comprised grid data with a 10 m × 10 m spatial resolution. Therefore, each area was calculated based on the number of graded cells clipped on the landslide hazard map.
2.4 SLA and Non-SLA altitude and slope analysis
We used ArcMap 10.8 and a digital topographic map (1:5,000) provided by the National Geographic Information Institute (Suwon, South Korea) to perform GIS analysis for determining the causes of slow-moving landslides. Altitude and slope differences between SLAs and non-SLAs were analyzed by generating a 5 m × 5 m DEM and slope map. The boundaries of each SLA and non-SLA in each study site were used to extract the altitude and slope data. Kolmogorov–Smirnov normality tests () were performed to determine the altitude and slope differences between SLAs and non-SLAs. This test was followed by a non-parametric Mann–Whitney U test (). Owing to the statistical reliability problems of the parametric test (t-test) of non-normally distributed data, a non-parametric test was performed.
We analyzed and compared the slopes of SLAs and non-SLAs to determine slope characteristics, which are expected to be one of the critical factors affecting slow-moving landslides on a portion of a mountain slope. The area data generated for SLA and non-SLA slopes were grouped into 5° intervals to analyze the differences in area. Kolmogorov–Smirnov and the non-parametric Kruskal–Wallis tests () were used to evaluate differences. The differences between SLA and non-SLA groups were ranked and assessed post hoc using Duncan’s multiple range test for non-parametric testing. The post hoc test of non-normally distributed data was performed after generating ranking data from the raw data ().
Meanwhile, this study focused on locations where slow-moving landslides had caused human or material damage in mountainous areas near human settlements or social infrastructure; we could not secure many samples. In other words, we had to perform the analysis using samples more minor than the general minimum sample size required for the data to be statistically normally distributed (df ≥ 30). To address this limitation, we performed a normality test on the sample data before each analysis to confirm that it was not normally distributed. We then used non-parametric tests to verify the statistical significance of the data, as these tests do not require the data to be normally distributed. Statistical analyses were performed using SPSS Ver. 21 (IBM SPSS, Armonk, NY, United States), with a significance level of p < 0.05.
2.5 Waterway analysis
We analyzed the waterways using the Arc Hydro Tools in ArcMap. Arc Hydro is a GIS tool that is used to analyze topographic and hydrological factors (). The Arc Hydro Model offers network tools for establishing a spatial database based on vector data. The network tools provide various functions, such as generating the upstream-to-downstream streamflow networks, nodes (river intersections), and reaches (linear watersheds linked to nodes) (). Based on the principle that water naturally flows from high-to low-altitude areas, the DEM can be used to determine the flow direction in a watershed. Arc Hydro Tools can be used to calculate water flow, mountain slope, and total rainfall in each region, and the numerical data can be used to analyze the stream basin ().
According to , the cause of slow-moving landslides is primarily affected by the groundwater level, as opposed to landslides, which are considerably more affected by the type or intensity of rainfall than by the groundwater level. Additionally, indented mountain areas are susceptible to slow-moving landslides because of the ease of groundwater inflow during rainfall (; ). Despite the importance of waterways created in mountainous areas by rainfall to the occurrence of slow-moving landslides, research has yet to be conducted on this topic. Therefore, we verified the roles of waterways as slow-moving landslide-triggering factors through a comparative analysis between SLAs and non-SLAs.
After estimating the direction of the waterways and rainfall catchments in SLAs and non-SLAs, we generated watersheds using a DEM created from a digital topographic map. The generated watersheds were used to delineate the waterways, and each waterway was extracted (; https://resources.arcgis.com/en/help/main/10.2/index.html#/An_overview_of_the_Hydrology_tools/009z0000004w000000). The differences in waterways between SLAs and non-SLAs were determined using Shapiro–Wilk normality test (for waterway length) followed by the non-parametric Mann–Whitney U test. Shapiro–Wilk test was developed to test the normality of a limited sample size (less than 50 cases).
3 Results and discussion
3.1 Landslide hazard map and slow-moving landslide site overlap
In Figure 2, the red polygons represent the boundaries of the slow-moving landslide sites, and an overlap analysis with the landslide hazard map was performed using these polygons. In addition, yellow and green polygons represent SLAs and non-SLAs, respectively, which were to analyze differences in altitude, slope, and waterway (Figure 2).
The landslide hazard map used in the overlap analysis is a thematic map created using the past total landslide occurrence information in South Korea (n = 2,000) and a logistic regression analysis to assign weights to nine factors, namely, forest type, diameter class, slope degree, slope aspect, slope length, slope curvature, parent rock, soil depth, and topographical wetness index, and distinguish landslide occurrence probabilities according to five grades: very high (Grade 1), high (Grade 2), low (Grade 3), significantly low (Grade 4), and very low (Grade 5) (; https://sansatai.forest.go.kr). The landslide hazard map and slow-moving landslide site boundaries were then overlapped to compare the differences between landslides and slow-moving landslides.
The proportion of land area occupied by slow-moving landslide sites (n = 10) on the landslide hazard map was 34% for Grade 3, 21% for Grade 2, and 20% for Grade 4. Regions that did not fall within the graded area accounted for 8.5% of the total area. Grade 3 accounted for ≥50% of the slow-moving landslide sites in sites 2 and 7, whereas Grade 5 accounted for ≥90% in site 3 (Table 1). In conclusion, we found that slow-moving landslides occurred in regions with low-hazard grades or regions that did not fall within the graded area (non-hazard areas) on the landslide hazard map. While also considering multiple regions that were outside the graded area, showed that a slow-moving landslide site corresponding to Grades 3–5 was the largest on the landslide hazard map, and this finding is consistent with our results.
TABLE 1
| Site no. | Grade of slow-moving landslide site on the landslide hazard map (m2) | Total (m2) | |||||
|---|---|---|---|---|---|---|---|
| Grade 1 | Grade 2 | Grade 3 | Grade 4 | Grade 5 | Excluded area | ||
| 1 | - | 14,300 (17.0) | 20,700 (24.6) | 12,900 (15.4) | 800 (1.0) | 35,200 (42.0) | 83,900 (100.0) |
| 2 | - | 1,900 (5.8) | 18,600 (56.7) | 8,600 (26.2) | 700 (2.1) | 3,000 (9.2) | 32,800 (100.0) |
| 3 | - | 600 (3.3) | 1,100 (6.0) | 100 (0.6) | 16,400 (90.1) | - | 12,800 (100.0) |
| 4 | - | - | 5,000 (42.0) | 5,500 (46.2) | 1,000 (8.4) | 400 (3.4) | 11,900 (100.0) |
| 5 | 15,300 (8.0) | 62,600 (32.7) | 45,700 (23.8) | 33,000 (17.2) | 35,100 (18.3) | - | 191,700 (100.0) |
| 6 | 2,000 (4.1) | 7,400 (15.2) | 16,700 (34.3) | 15,500 (31.8) | 7,100 (14.6) | - | 48,700 (100.0) |
| 7 | - | 12,800 (13.5) | 48,500 (50.9) | 27,200 (28.6) | 3,800 (4.0) | 2,900 (3.1) | 95,200 (100.0) |
| 8 | - | 1,400 (6.4) | 9,200 (41.8) | 4,600 (20.9) | - | 6,800 (30.9) | 22,000 (100.0) |
| 9 | 300 (1.5) | 5,300 (26.8) | 9,400 (47.5) | 4,200 (21.2) | 600 (3.0) | - | 19,800 (100.0) |
| 10 | 4,700 (10.2) | 14,000 (30.2) | 21,400 (46.2) | 5,300 (11.5) | 700 (1.5) | 200 (0.4) | 46,300 (100.0) |
| Total (m2) | 22,300 (3.9) | 120,300 (21.1) | 196,300 (34.4) | 116,900 (20.5) | 66,200 (11.6) | 48,500 (8.5) | 570,500 (100.0) |
Overlap results of the landslide hazard map and slow-moving landslide sites.
The values within parentheses represent the proportions of the area occupied by a slow-moving landslide site.
Landslide hazard maps are helpful indicators of landslide hazard areas, but they have limitations in predicting slow-moving landslide hazards. Landslides and slow-moving landslides are both disasters that cause ground collapse, but their causes are known to be different. Landslides are often caused by natural factors such as heavy rain or earthquakes, mainly on steep slopes (). On the other hand, Slow-moving landslides are often caused by excessive soil extraction (artificial forest development) () or rising groundwater levels (; ) and occur in areas with relatively gentle slopes (). Our research results show that many slow-moving landslide sites are located on the landslide hazard map in low-hazard or non-hazard areas. This result confirms that slow-moving landslides are caused by factors different from those that cause landslides. The differences between slow-moving landslides and landslides can also be confirmed by comparing the characteristics and phenomena of their causative factors, as reported by the National Institute of Forest Science in South Korea (; Table 2).
TABLE 2
| Classification | Slow-moving landslide | Landslide |
|---|---|---|
| Geological features | Occurs frequently in specific geological features or geological structures | Low correlation with geological features |
| Soil | Slip surface of cohesive soil | Surface soil of sandy soil or decomposed granite soil |
| Topography | Flat top with a gentle slope of 5°–20° | The valley head or top of the slope with a steep slope of over 20° |
| Slope | 10°–20° | 35°–60° |
| Activity status | Persistent, recurrent, and highly time-dependent | The suddenness and low time-dependent |
| Movement status | Usually slow (mostly 0.01–10 mm/day) | Very fast (more than 10 mm/day) |
| Lump of earth | Often moves while maintaining its original state | Disturbed |
| Cause | Underground water | Rainfall (especially rainfall intensity) |
| Occurrence scale | Large scale from 1 to 100 ha | On average, around 5 ha |
| Signal | Cracks, depressions, uplifts, and fluctuations of underground water level | Few signals and sudden |
Comparison of the triggering factor characteristics and phenomena between slow-moving landslides and landslides in South Korea.
Slow-moving landslides in South Korea wereconsidered a type of landslide; however, slow-moving landslides are characterized by a gradual movement of the soil stratum, whereas landslides are a type of superficial stratum collapse followed by rapid, often large-scale, mountain erosion (; ). Given the damage caused by slow-moving landslides, they should be regarded as distinct phenomena in order to establish a national slow-moving landslide hazard map with similar predictive capabilities as landslide hazard maps.
3.2 Differences in altitude
We generated altitude data for the SLAs and non-SLAs (See Figure 2) to analyze the differences between areas. The average altitude was 213.9 m in SLAs and 209.0 m in non-SLAs. The results of Kolmogorov–Smirnov normality test showed that the altitude data for the two regions were not normally distributed (Table 3).
TABLE 3
| Classification | Mean (m) | SD (m) | Skewness | Kurtosis | Kolmogorov–Smirnov | ||
|---|---|---|---|---|---|---|---|
| Statistic | df | p | |||||
| SLAs | 213.94 | 0.551 | 0.426 | −0.945 | 0.076 | 30,919 | <0.001 |
| Non-SLAs | 209.03 | 8.9 | 11.550 | 198.893 | 0.173 | 30,861 | <0.001 |
Normality test results for altitude data of SLAs and non-SLAs.
The non-parametric Mann–Whitney U test showed that the altitude of SLAs and non-SLAs differed significantly (p < 0.001; Table 4). Therefore, slow-moving landslides occurred at relatively higher altitudes than those in the surrounding mountains.
TABLE 4
| Classification | No. of cell (no.) | Rank of mean | P |
|---|---|---|---|
| SLAs | 30,919 | 32,068.76 | <0.001 |
| Non-SLAs | 30,861 | 29,710.03 | |
| Total | 61,780 | - |
Non-parametric Mann–Whitney U test results for altitude differences between SLAs and non-SLAs.
It was determined that mountain slopes with relatively low altitudes collect a lot of surface water due to rainfall or heavy rain, and this surface water infiltrates underground again, raising the groundwater level on the slopes (; ). Rising groundwater levels increase ground pressure (; ; ; ), causing slow-moving landslides. These results suggest that the altitude characteristics of mountain slopes should be considered when predicting and preventing slow-moving landslides.
3.3 Differences in slope
Slope data were generated for the SLAs (yellow polygons) and non-SLAs (green polygons) (Figure 3) and analyzed to determine the differences between areas.
FIGURE 3
The average SLA and non-SLA slopes were 22.1° and 21.1°, respectively. The results of Kolmogorov–Smirnov test showed that the slope data were not normally distributed (Table 5).
TABLE 5
| Classification | Mean (°) | SD (°) | Skewness | Kurtosis | Kolmogorov–Smirnov | ||
|---|---|---|---|---|---|---|---|
| Statistics | df | p | |||||
| SLAs | 22.1 | 7.9 | −0.269 | −0.001 | 0.057 | 30,913 | <0.001 |
| Non-SLAs | 21.1 | 8.9 | −0.066 | −0.192 | 0.049 | 30,763 | <0.001 |
Normality test results for the slope data of SLAs and non-SLAs.
The non-parametric Mann–Whitney U test showed that the slope differed significantly between SLAs and non-SLAs (p < 0.001; Table 6). Therefore, slow-moving landslides occurred in areas with steeper slopes than those in the surrounding mountains.
TABLE 6
| Classification | No. of cell (no.) | Rank of mean | P |
|---|---|---|---|
| SLAs | 30,913 | 31,920.67 | <0.001 |
| Non-SLAs | 30,763 | 29,751.06 | |
| Total | 61,676 | - |
Non-parametric Mann–Whitney U test results for slope differences between SLAs and non-SLAs.
These results suggest that slope characteristics must be considered when predicting slow-moving landslides in mountain areas. We must not overlook the fact that artificial factors such as damage to mountain areas or groundwater development can affect the slope of stable mountain areas, increasing the risk of slow-moving landslides.
We analyzed the slope characteristics of SLAs and non-SLAs based on area data. Kolmogorov–Smirnov normality test showed that the slope area data were not normally distributed (Table 7).
TABLE 7
| Classification | Mean (m2) | SD (m2) | Skewness | Kurtosis | Kolmogorov–Smirnov | ||
|---|---|---|---|---|---|---|---|
| Statistics | df | p | |||||
| SLAs | 9,378.1 | 15,071.0 | 2.993 | 10.029 | 0.267 | 82 | <0.001 |
| Non-SLAs | 8,866.1 | 13,359.2 | 2.710 | 8.405 | 0.254 | 87 | <0.001 |
Normality test results for the slope area data of SLAs and non-SLAs.
The results of the non-parametric Kruskal–Wallis test showed that the area for each slope differed significantly in the SLAs (p < 0.001). An analysis of variance (ANOVA) was used to assess the differences between slope groups based on the ranking data of the area of each slope, and the results showed that the groups differed significantly (p < 0.001). In the post hoc analysis, the 45° group displayed a low rank, whereas the 25° group presented a high rank (Table 8). Therefore, slow-moving landslides occurred in vast areas with 21°–25° slopes.
TABLE 8
| Slope group (°) | Area (m2) | p (Kruskal–Wallis) | Rank of slope area | ||||
|---|---|---|---|---|---|---|---|
| N | Meanz | SD | F | p | |||
| 5 | 19,550 | <0.001 | 9 | 23.22ab | 16.16 | 5.222 | <0.001 |
| 10 | 34,525 | 9 | 31.33bc | 18.96 | |||
| 15 | 102,575 | 10 | 41.55bcde | 21.85 | |||
| 20 | 145,350 | 10 | 53.20de | 18.91 | |||
| 25 | 192,400 | 10 | 59.80e | 15.82 | |||
| 30 | 165,900 | 10 | 51.25cde | 20.32 | |||
| 35 | 83,525 | 9 | 38.39bcd | 26.63 | |||
| 40 | 21,950 | 6 | 32.67bcd | 16.31 | |||
| 45 | 2,575 | 6 | 11.58a | 7.39 | |||
| Total | 768,350 | 79 | 40.00 | 25.00 | |||
Differences in the area occupied by slope groups in SLAs.
zDuncan’s multiple range test, p < 0.05.
showed that slow-moving landslides typically occurred on gentle slopes of 5°–20°, and claimed that the average slope of slow-moving landslide occurrence areas was 23.8°. In the present study, the largest area of slow-moving landslide occurrence was found on 21°–25° slopes. Research in Japan and South Korea has solely relied on a few events to establish the range of the slope factor in the assessment of slopes as triggering factors of previous slow-moving landslides. However, because of the difference between the results of this study and those of previous studies, the range of the slope factor should be re-established.
The area for each slope differed significantly in the non-SLAs (p < 0.001). The ANOVA based on the ranking data of slope area showed that the slope groups differed significantly (p < 0.001). In particular, the 50° group displayed a low rank, whereas the 25° group presented a high rank (Table 9).
TABLE 9
| Slope group (°) | Area (m2) | p (Kruskal–Wallis) | Rank of slope area | ||||
|---|---|---|---|---|---|---|---|
| N | Meanz | SD | F | p | |||
| 5 | 38,750 | <0.001 | 9 | 25.06ab | 24.07 | 4.801 | <0.001 |
| 10 | 58,525 | 10 | 34.65bc | 22.81 | |||
| 15 | 109,000 | 10 | 55.75cd | 22.06 | |||
| 20 | 141,475 | 10 | 57.50cd | 20.47 | |||
| 25 | 176,725 | 10 | 62.20d | 19.23 | |||
| 30 | 141,800 | 10 | 55.10cd | 22.23 | |||
| 35 | 68,375 | 10 | 41.60bcd | 24.88 | |||
| 40 | 25,850 | 7 | 38.64bcd | 16.66 | |||
| 45 | 8,975 | 7 | 20.64ab | 14.14 | |||
| 50 | 1,450 | 3 | 10.83a | 7.22 | |||
| Total | 770,925 | 86 | 43.50 | 25.00 | |||
Differences in the areas of slope groups in non-SLAs.
zDuncan’s multiple range test, p < 0.05.
Analysis of the slope area difference between the SLAs and non-SLAs revealed that the 25° group had the largest area. This finding indicated that slow-moving landslides did not occur in non-SLAs, despite a sufficiently high probability of their occurrence in the future. Slow-moving landslides are not necessarily isolated events; they can also be continuous and recurrent, which necessitates the long-term analysis and management of slow-moving landslide sites and their surroundings. Site 1 is a critical case in this study. This region reportedly experienced a slow-moving landslide in August 2002. However, four past instances of slow-moving landslides widened the restoration area and caused considerable loss of life and property damage (; ; ; ).
3.4 Differences in waterway characteristics
Waterway data were generated for the SLAs (blue lines) and non-SLAs (pink lines) (Figure 4), and the differences in waterway length between areas were analyzed using the Arc Hydro Tools. Shapiro–Wilk normality test showed that the waterway data were not normally distributed (Table 10).
FIGURE 4
TABLE 10
| Classification | Mean (m) | SD (m) | Skewness | Kurtosis | Shapiro–Wilk | ||
|---|---|---|---|---|---|---|---|
| Statistics | df | p | |||||
| SLAs | 85.2 | 154.3 | 2.812 | 8.334 | 0.584 | 10 | <0.001 |
| Non-SLAs | 205.4 | 221.2 | 2.334 | 5.951 | 0.715 | 10 | 0.001 |
Normality test results for waterway data of SLAs and non-SLAs.
The non-parametric Mann–Whitney U test showed that the average rank of SLA waterway length was 7.3 (average 85.2 m) and that of non-SLA waterway length was 13.7 (average 205.4 m), indicating that non-SLA waterways were longer (p < 0.05; Table 11).
TABLE 11
| Classification | N | Rank of mean | p |
|---|---|---|---|
| SLAs | 10 | 7.3 | 0.015 |
| Non-SLAs | 10 | 13.7 | |
| Total | 20 | - |
Non-parametric Mann–Whitney U test results for differences in waterway length between SLAs and non-SLAs.
Based on our findings, we determined that the hazard of slow-moving landslides is high in the inner area where a long waterway is formed around the slope of a mountainous area. In conclusion, groundwater-related factors (; ), as well as novel slow-moving landslide-triggering factors, such as waterway characteristics, are essential for national-level slow-moving landslide vulnerability assessments. Meanwhile, this study used GIS to focus on the length of waterways created within mountainous areas. However, there is a need to further develop the characteristics of waterway factors by comprehensively analyzing not only the length of the created waterway but also factors such as the width and depth of the waterway. In addition, the results of this research methodology can be used to predict the occurrence of gully erosion through local comparison.
Slow-moving landslides can be caused by various factors, including the altitude, slope, and waterway characteristics in this research and the geological features of mountain slopes, rainfall, groundwater, reservoir water level, and the composition of forests, as confirmed by previous studies. Before comprehensively analyzing the impact of these various factors, we first analyzed the differences between SLAs and non-SLAs in terms of topographic characteristics such as altitude and slope and waterways generated by topographic features and rainfall. And we confirmed that these differences were statistically significant. This suggests that altitude, slope, and waterway characteristics are still characteristic factors causing slow-moving landslides. However, it is undeniable that comprehensive research that considers not only altitude, slope, and waterways but also the geological characteristics of mountain slopes, the composition of forests, land cover, as well as meteorological factors such as drought, rainfall, and snowfall, is necessary for predicting the hazard of slow-moving landslides. This will allow us to quickly predict, preemptively manage, and prepare preventive measures in areas in danger of slow-moving landslides.
4 Conclusion
Slow-moving landslides are caused by a combination of internal and external factors. At times, slow-moving landslides can accelerate rapidly and collapse catastrophically (). In contrast to landslides, slow-moving landslides, which were not actively investigated until recently, require further research. The development of an early warning system at the national level might enable us to minimize damage to social infrastructure and local communities caused by slow-moving landslides.
Slow-moving landslides can cause damage to a wide range of infrastructure, including forests, agricultural fields, roads, and buildings, and can also cause fatalities. Therefore, research on the factors causing slow-moving landslides is essential for predicting the hazard of slow-moving landslides and preparing preventive measures. Accordingly, we scientifically verified the topographic triggers of slow-moving landslides in this study and explored a new trigger factor called waterway. This study is expected to help researchers narrow the scope of verification for the many factors that cause natural disasters in mountain areas, such as slow-moving landslides.
In this study, GIS was used to verify the differences between landslide and slow-moving landslide occurrences by overlapping a landslide hazard map and slow-moving landslide sites. SLAs and non-SLAs were designated based on slow-moving landslide site boundaries to compare their altitude, slope, and waterway characteristic differences.
Grade 3 slow-moving landslide sites accounted for the largest area on the landslide hazard map, even when non-hazard areas were considered. Slow-moving landslide occurrences clearly differed from landslides, emphasizing the need for a distinct national slow-moving landslide hazard information system. In a comparative analysis of the altitude and slope of SLAs and non-SLAs, the SLAs were found to have higher altitudes and steeper slopes than the non-SLAs. The 25° (21°–25°) group comprised the largest SLA and non-SLA areas. This result suggests that slow-moving landslides are likely to occur in the surrounding non-SLAs. We analyzed the differences in slope area for the SLAs and non-SLAs and discovered that ≥35° slopes, which were associated with the highest number of landslides, had the lowest occurrence rate in SLAs. Waterways created by precipitation and torrential rains were longer in non-SLAs than in SLAs. Therefore, waterways could be a novel triggering factor for slow-moving landslides. The generated water could seep into the ground and converge in a certain location, increasing the hazard of slow-moving landslides. Additional research is needed to investigate the impact of waterways on groundwater cycles and slow-moving landslide hazards.
By scientifically analyzing the characteristics of slow-moving landslide-causing factors in this study, we can better understand their causes and develop technology to predict them. In particular, the factors in this study can be used to construct a slow-moving landslide hazard prediction model to identify areas in danger of slow-moving landslides. This is expected to lead to more accurate predictions of slow-moving landslide hazards by establishing a national slow-moving landslide hazard information system, enabling effective damage prevention measures to be prepared.
The Korea Forest Service aims to develop a slow-moving landslide hazard map by 2024. To promote the widespread management of slow-moving landslides and reduce fatalities and property damage, regions with topographies similar to South Korea, such as China and Japan, would benefit from slow-moving landslide hazard maps and information systems that incorporate various known and novel factors, such as the geology and topography of mountainous areas as well as altitude, slope, and waterway characteristics. Based on the findings of this study, future studies should investigate the factors that cause slow-moving landslides, such as maximum slope and area or volume of waterways. At this point, by assigning the shapes of SLAs and non-SLAs to various forms, such as circles or squares, it is possible to establish the reliability of trigger factors by comparing them. In addition, a study on the geology or topographical characteristics (factors used to create a landslide hazard map) of slow-moving landslide sites should be conducted, which was not covered in this study. For the findings of this study to be universally applicable, further information on slow-moving landslide sites should be obtained and analyzed in the future and compared with the results of this study. In addition, weights derived from correlation or regression analyses between factors should be applied to each factor in order to develop a reliable national slow-moving landslide hazard information system and map.
Statements
Data availability statement
The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.
Author contributions
G-MT: Conceptualization, Data curation, Formal Analysis, Methodology, Resources, Software, Validation, Visualization, Writing–original draft, Writing–review and editing. J-HP: Conceptualization, Funding acquisition, Investigation, Project administration, Resources, Supervision, Validation, Writing–review and editing. C-KL: Conceptualization, Validation, Writing–review and editing. HK: Data curation, Formal Analysis, Methodology, Resources, Software, Supervision, Validation, Visualization, Writing–original draft, Writing–review and editing, Conceptualization.
Funding
The author(s) declare financial support was received for the research, authorship, and/or publication of this article. This study was conducted with the support of the R&D Program for Forest Science Technology (Project No. 2021347B10-2123-CD01) funded by the Korea Forest Service (Korea Forestry Promotion Institute, Seoul, South Korea).
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1
AgliardiF.ScuderiM. M.FusiN.CollettiniC. (2020). Slow-to-fast transition of giant creeping rockslides modulated by undrained loading in basal shear zones. Nat. Commun.11, 1352. 10.1038/s41467-020-15093-3
2
An Overview of the Hydrology Toolset (2023). An overview of the Hydrology toolset. Available at: https://resources.arcgis.com/en/help/main/10.2/index.html#/An_overview_of_the_Hydrology_tools/009z0000004w000000 (Accessed August 8, 2023).
3
ChoiJ.ChoiB.KimN.LeeC.SeoJ.JeonB. (2021). Estimation of potential risk and numerical simulations of landslide disaster based on UAV photogrammetry. KSCE J. Civ. Environ. Eng. Res.41, 675–686. 10.12652/Ksce.2021.41.6.0675
4
ChoiJ.EomK.JeonK.YoonY. (2019). Monitoring ground movements for deep seated landslide risk areas during an earthquake episode Regular Conferences of Korean Society of Civil Engineers2019, 960–961.
5
DasG.LepchaK. (2019). Application of logistic regression (LR) and frequency ratio (FR) models for landslide susceptibility mapping in Relli Khola River Basin of Darjeeling Himalaya, India. SN Appl. Sci.1, 1453. 10.1007/s42452-019-1499-8
6
HabumugishaJ. M.ChenN.RahmanM.IslamM. M.AhmadH.ElbeltagiA.et al (2022). Landslide susceptibility mapping with deep learning algorithms. Sustainability14, 1734. 10.3390/su14031734
7
HamD.HwangS. (2014). Review of landslide forecast standard suitability by analysing landslide-inducing rainfall. J. Korean Soc. Hazard Mitig.14, 299–310. 10.9798/KOSHAM.2014.14.3.299
8
HighlandL. M.BobrowskyP. (2008). The landslide handbook-A guide to understanding landslides. Reston, Virginia: U.S. Geological Survey Circular,1325, 4–26.
9
JauJ.ParkS.SonD.JooS. (2000). The effects of geological and topographical features on landslide and land-creep. J. Korean Soc. For. Sci.89, 323–334.
10
JeonB.LeeS. (2019). A study on the effect of collector well on the landcreep slope. J. Eng. Geol.29, 123–136. 10.9720/kseg.2019.2.123
11
JeonG. (2011). Soil conservation engineering. Seoul, South Korea: Hangmoonsa, 103–105.
12
JeongJ. (2013). SPSS statistics descriptive statistics and correlation analysis. 12th ed. Seoul South Korea: Data Solution, 29–39.
13
KangX.WangS.WuW.XuG.ZhaoJ.LiuF. (2022). Soil-water interaction affecting a deep-seated landslide: from field monitoring to experimental analysis. Bull. Eng. Geol. Environ.81, 82. 10.1007/s10064-021-02556-0
14
KimH.KimM.LeeM.ParkY.KwakJ. (2017). Correlation of deep landslide occurrence and variation of groundwater level. J. Korea Soc. For. Eng.15, 1–12.
15
KimJ.KimB.JeongH.ShinS. (2018). Environmental statistics and data analysis. Seoul, South Korea: Hannarae Publishing Co., 158–173.
16
KimJ.KimJ.LeeS.ChoK.KimJ. (2021). Interpretation of electrical resistivity tomogram with contents of clay minerals for the land creeping area. J. Eng. Geol.31, 187–197. 10.9720/kseg.2021.2.187
17
KimK.ChoiY.KimJ.KohJ. T.ChunB. J.BaeC. S.et al (2004). The promoter of brain-specific angiogenesis inhibitor 1-associated protein 4 drives developmentally targeted transgene expression mainly in adult cerebral cortex and hippocampus. J. Korean Assoc. Geogr. Inf. Stud.7, 87–94. 10.1016/j.febslet.2004.03.106
18
KimK.KangM.ParkJ. (2016). Experimental research for collapse of tension crack in land creeping area. Hadong. J. Agric. Life Sci.50, 11–17. 10.14397/jals.2016.50.6.11
19
KimN.ChoiB.ChoiJ.JeonB. (2020). Time series analysis of soil creep on cut slopes using unmanned aerial photogrammetry. J. Eng. Geol.31, 447–456. 10.9720/kseg.2020.4.447
20
KimS.KimD.MaengS. (2012). A risk analysis of water courses and landslide using contour maps: focusing on Mt. Seonggo in cheonan city. J. Digit. Converg.10, 289–296.
21
Korea Forest Service (2021). National landslide prevention plan. Seoul, South Korea: Korea Forest Service, 2–4.
22
LacroixP.HandwergerA. L.BièvreG. (2020). Life and death of slow-moving landslides. Nat. Rev. Earth Environ.1, 404–419. 10.1038/s43017-020-0072-8
23
Landslide Information System: About Landslide Hazard Map (2023). sansatai. Available at: https://sansatai.forest.go.kr (Accessed August 8, 2023).
24
LeeC.SeoJ.WooC.EuS. (2022).Development of a novel forest soil sediment disaster damage reduction technology: A focus on slow-moving landslides and earthquake-induced forest soil sediment disasters. Seoul, South Korea: National Institute of Forest Science, 8.
25
LeeC.WooC.SeoJ.EuS. (2021). Understanding landslides properly. Seoul, South Korea: National Institute of Forest Science, 7–14.
26
LeeM.ParkJ.ParkY. (2019). Analysis of characteristics using geotechnical investigation on the slow-moving landslide in the Pohang-Si area. J. Korean Soc. For. Sci.108, 233–240. 10.14578/jkfs.2019.108.2.233
27
LeeS.KimJ.KimK.KwonI. (2020). Delineation of the slip weak zone of land creeping with integrated geophysical methods and slope stability analysis. J. Eng. Geol.30, 289–302. 10.9720/kseg.2020.3.289
28
National Institute of Forest Science (2021). 15 things needed to be known about land creeps in a Nutshell. Seoul, South Korea: National Institute of Forest Science, 4–6.
29
ParkJ. (2015). Analysis on the characteristics of the landslide: with a special reference on geo-topographical characteristics. J. Korean Soc. For. Sci.104, 588–597. 10.14578/jkfs.2015.104.4.588
30
ParkJ.ChoiK.LeeS.MaH.LeeJ.WooB. (2003a). Analysis on the characteristics of the landslide in nasamri (I): with a special reference on geo-topographical characteristics. J. Korean Soc. For. Sci.92, 246–253.
31
ParkJ.ChoiK.LeeS.MaH.LeeJ.WooB. (2003b). Analysis on the characteristics of the landslide in nasamri (II): with a special reference on cause of landslide. J. Korean Soc. For. Sci.92, 526–535.
32
ParkJ.KimS.LeeS.KangH. (2022). Study on the characteristics of the slow-moving landslide (land creep) in the Sanji Valley of Jinju. J. Korean Soc. For. Sci.111, 115–124. 10.14578/jkfs.2022.111.1.115
33
ParkJ.LeeS.ChoiK.MaH.LeeJ.WooB. (2004). Analysis on the characteristics of the landslide in nasamri (III): with a special reference on stability analysis. J. Korean Soc. For. Sci.93, 188–197.
34
ParkJ.MaH.ParkS. (2021a). Actual conditions on the land creep areas by the analysis of satellite imagery. J. Agric. Life Sci.55, 57–64. 10.14397/jals.2021.55.1.57
35
ParkJ.ParkS. (2018). Analysis of instances of characteristics land creep on the mine area in Korea. J. Korean Soc. For. Sci.107, 393–401. 10.14578/jkfs.2018.107.4.393
36
ParkJ.ParkS.DavidsonE.De VeraM. A. (2020). Insights on mental health when living with rheumatoid arthritis: A descriptive qualitative study of threads on the reddit website. J. Korean Soc. For. Sci.109, 62–71. 10.1186/s41927-020-00163-2
37
ParkJ.ParkS.KangM.ImH. (2021b). Analysis on scale and geological characteristics-based influence factors in land creep. J. Agric. Life Sci.55, 37–51. 10.14397/jals.2021.55.5.37
38
ParkJ.SeoJ.LeeC. (2019). Analysis of GIS for characteristics on the slow-moving landslide: with a special reference on slope and grade of landslide. J. Korean Soc. For. Sci.108, 311–321. 10.14578/jkfs.2019.108.3.311
39
PreisigG.EberhardtE.SmithymanM.AlexanderP.BonzanigoL. (2016). Hydromechanical rock mass fatigue in deep-seated landslides accompanying seasonal variations in pore pressures. Rock Mech. Rock Eng.49, 2333–2351. 10.1007/s00603-016-0912-5
40
SenanayakeA.FernandoN.WasanaM.AmaratungaD.HaighR.MalalgodaC.et al (2022). Landslide induced displacement and relocation options: A case study of owner driven settings in Sri Lanka. Sustainability14, 1906. 10.3390/su14031906
41
SeoY.KimS.LeeK. (2008). 3-D slope stability analysis on groundwater level changes in Oksan landslide area. J. Eng. Geol.18, 167–173.
42
ShinH.KimM.WoonW. (2018). Application of geophysical exploration technique to the identification of the active weak zones in large scale mountainous region. Geophys. Geophys. Explor.21, 162–170. 10.7582/GGE.2018.21.3.162
43
StanleyT.KirschbaumD. B. (2017). A heuristic approach to global landslide susceptibility mapping. Nat. Hazards87, 145–164. 10.1007/s11069-017-2757-y
44
TianS.HuG.ChenN.RahmanM.NiH.Somos-ValenzuelaM. (2023). Effects of tectonic setting and hydraulic properties on silent large-scale landslides: A case study of the zhaobishan landslide, China. Int. J. Disaster Risk Sci.14, 600–617. 10.1007/s13753-023-00502-0
45
WangJ.WangS.SuA.XiangW.XiongC.BlumO. (2021). Simulating landslide-induced tsunamis in the yangtze river at the three gorges in China. Acta Geotech.16, 2487–2503. 10.1007/s11440-020-01131-3
46
WooB.ParkJ.ChoiH.JeonG.KimK. (1996). A study on the characteristics of the landslide in hyuseok-dong (I): topographical characteristics and surface displacement. J. Korean Soc. For. Sci.85, 565–570.
47
WooS. (2010). Pasw: statistics 18.0. Seoul, South Korea: Human and Welfare, 383.
48
YouH.KimD.YangS. (2012). Development of GIS-based method for estimating and representing stream slopes along the river network. J. Environ. Sci. Int.21, 725–738. 10.5322/JES.2012.21.6.725
49
YuP.LiuH.YuH.XieY.YuY.ZhuC.et al (2023). Study on fluid-solid coupling numerical simulation and early warning of weathered granite landslides induced by extreme rainfall. Sustainability15, 11738. 10.3390/su151511738
50
ZhangZ.ZengR.MengX.ZhaoS.WangS.MaJ.et al (2023). Effects of changes in soil properties caused by progressive infiltration of rainwater on rainfall-induced landslides. Catena.233, 107475. 10.1016/j.catena.2023.107475
Summary
Keywords
land creep, landslide, slow-moving landslide, triggering factor, waterway
Citation
Tak G-M, Park J-H, Lee C-K and Kim H (2023) Influence of altitude, slope, and waterway characteristics on the occurrence of slow-moving landslides in South Korea. Front. Earth Sci. 11:1276768. doi: 10.3389/feart.2023.1276768
Received
15 August 2023
Accepted
26 September 2023
Published
09 October 2023
Volume
11 - 2023
Edited by
Chong Xu, University of Ferrara, Italy
Reviewed by
Liye Feng, Ministry of Emergency Management, China
Zhilu Chang, Nanchang University, China
Ionut Cristi Nicu, Norwegian Institute for Cultural Heritage Research, Norway
Gaetano Robustelli, University of Calabria, Italy
Updates
Copyright
© 2023 Tak, Park, Lee and Kim.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Hyun Kim, kimhyun@gnu.ac.kr
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.