1 Introduction
Landslide relics are topographic features and geomorphic markers where landslides have occurred in geological history. They can provide important information about past landslide events and help analyze the mechanisms, frequency, and scale of landslides in the region (
Malamud et al., 2004;
Guzzetti et al., 2012;
Li et al., 2022;
Luetzenburg et al., 2022;
Huang et al., 2023). In recent years, with the widespread application of artificial intelligence algorithms, such as big data, neural networks, and machine learning, in the analysis and evaluation of landslide mechanisms (
He et al., 2024), the construction of a high-precision, all-factor landslide sample database has become a crucial step in enhancing the reliability of geological disaster prediction models (
Guzzetti et al., 2012;
Gerzsenyi and Albert, 2021;
Huang et al., 2024b;
Shao et al., 2024).
Nyingchi City is located in the south-east of the Xizang Autonomous Region, China. This area has an extremely special natural geographical environment. It is an important transportation hub in Xizang, second only to Lhasa, and also serves as a construction site for numerous large-scale infrastructure projects in China (
Zhao et al., 2024a). Important transportation routes, such as the Sichuan-Xizang Railway and National Highway 318, pass through Nyingchi (
Yang et al., 2023). While unlocking the local tourism resources, they have also intensified the damage caused by human activities to the stability of the regional geological environment. Under the combined influence of multiple factors, including meteorology, topography, active faults, climate change, and human activities, the Nyingchi area is subject to strong internal and external dynamic processes, characterized by a fragile geological environment, frequent geological disasters, and prominent disaster chain characteristics. This poses a serious threat to people’s lives and property, as well as to the safety of engineering construction and the regular operation of critical infrastructure (
Liu et al., 2022;
Wang et al., 2024;
Yu et al., 2024).
Currently, many scholars have been researching the identification of landslide relics in the Nyingchi area. Most of these works are conducted along the Sichuan-Xizang Railway, such as the landslide relics inventory of Gyaca-Nang County (
Guo et al., 2019), the landslide relics inventory of Ya’an-Nyingchi (
Wang et al., 2022c), and the landslide relics inventory of Qamdo-Nyingchi (
Zhao et al., 2023). In addition, some scholars have chosen a broader area to interpret landslide relics. For example,
Jia et al. (2021) took the eastern part of Nyingchi as the target area and established an inventory of landslide relics containing 1766 samples;
Zang et al. (2024) counted the landslides in the south-eastern part of the Qinghai-Xizang Plateau since 2009 and established a landslide inventory with 659 samples in combination with fieldwork. Overall, although multiple landslide-related databases have been compiled in Nyingchi, a systematic landslide relic inventory covering the entire Nyingchi administrative region is still lacking, particularly in remote areas far from human settlements and major engineering facilities, where a clear gap exists in the identification of landslide relics. Related studies have shown that evaluating the susceptibility of regional landslides based on local landslide data can lead to biased results (
Huang et al., 2024a). This issue not only reduces the accuracy of regional-scale geological hazard risk assessment in Nyingchi but also leads to the inability to provide accurate data support and decision-making reference for the local government to formulate scientific and comprehensive geological hazard risk prevention and control strategies, which restricts the promotion of regional geological hazard prevention and control work.
In view of this, this paper takes the entire area of Nyingchi as the research scope, uses high-resolution remote sensing images combined with DEM data to systematically interpret landslides with sliding traces in the region, and constructs an inventory of landslide relics covering the entire Nyingchi. At the same time, combining existing landslide records to conduct multi-source data verification and content supplementation on the compiled inventory further improves the reliability of the data. On this basis, this study also conducted a preliminary analysis of the spatial distribution characteristics of landslide relics. The research results will provide a solid data foundation for regional landslide risk prevention and control, as well as national spatial planning in Nyingchi, thereby providing strong support for ensuring the smooth construction and operation of major projects in the region.
2 Geological background
The Nyingchi area is situated in the south-east of the Xizang Autonomous Region, covering an approximate total area of 150281 km
2. It comprises seven counties and districts, namely Medog County, Zayu County, Gongbo’gyamda County, Bayi District, Nang County, Bomi County, and Mainling County, with a permanent population of about 230000. To the north of Nyingchi City lie the Nyainqêntanglha Mountains; to the south, the eastern section of the Himalayas; to the north-west, the remnants of the Gangdise Mountains; and to the east, the Hengduan Mountains. The overall terrain is high in the north and low in the south, with a sharp contrast in altitude within the region. It is home to the world’s deepest canyon, the Yarlung Zangbo Grand Canyon, as well as the highest peak in the eastern Himalayas, Mount Namcha Barwa, making it a typical alpine canyon area (
Zhao et al., 2019;
Guo et al., 2024). Controlled by complex topographic features, meteorological factors such as temperature and rainfall in the study area exhibit significant spatial and temporal differences, resulting in distinct climate types, including tropical, subtropical, temperate, and frigid zones, with both humid and semi-humid climates coexisting.
Geologically, the study area lies at the eastern tectonic junction of the Qinghai-Xizang Plateau, at the core of the collision zone between the Indian and Eurasian Plates (
Tapponnier et al., 1982). It is one of the regions with the highest tectonic uplift and erosion in the world (
Li et al., 2018). Complex orogenic belts and deep and large faults characterize its tectonic background. The main fault zones in the area include the Himalayan Fault Zone, Medog Fault Zone, Aparon Fault Zone, Yarlung Zangbo River Fault Zone, and Lijia-Zayu Fault Zone (
Wu et al., 2021). Controlled by active faults, the study area has intense seismic activity. Historically, there have been 19 earthquakes with a magnitude of
Ms 6.0 or greater. The largest earthquake was the Ms 8.6 earthquake in Zayu in 1950, the strongest in China in the 20th century, with epicentral intensity reaching XII. It caused nearly 1500 deaths and enormous damage (
Ben-Menahem et al., 1974). In recent years, the largest earthquake in the Nyingchi area was the Ms 6.9 earthquake in Mainling on November 18, 2017, with the maximum intensity reaching degree VII (
Gao et al., 2025; Fig. 1).
3 Data and methods
3.1 Data
This paper interprets landslide relics using Google Earth, which contains a large number of high-resolution optical satellite images. The Google Earth platform includes various optical imagery, including SPOT 5 (2.5 m), WorldView-1 (0.5 m), WorldView-2 (0.5 m), and WorldView-3 (0.3 m).
Upon completing the landslide relic inventory for Nyingchi, we examined the spatial distribution characteristics of these features. Based on a synthesis of previous research on landslide disasters in the region, nine environmental factors were selected to analyze their spatial coupling with landslide relic distribution: elevation, slope aspect, slope gradient, land cover type, distance to rivers, distance to active faults, stratigraphy, mean annual temperature, and mean annual rainfall. The data sources for these factors are listed in Table 1, and their spatial distributions are illustrated in Fig. 2.
3.2 Methods
The method adopted in this paper is the human-computer interactive visual interpretation method, which involves visually observing and analyzing remote sensing images and delineating landslide relics on the Google Earth platform. After initially constructing the inventory of landslide relics, it is sampled and inspected by reviewing the literature and combining it with field surveys. The identified landslide relics are then supplemented and revised to ensure the accuracy of the final inventory. The workflow for constructing the entire inventory of landslide relics is shown in Fig. 3.
The human-computer interaction visual interpretation method requires interpreters to possess extensive professional knowledge and a high sensitivity to differences in the surrounding geological environment, as depicted in remote-sensing satellite images. These differences specifically include five aspects: topographic anomalies, vegetation anomalies, hydrological anomalies, accumulation anomalies, and image texture abnormality. The specific characteristics of various anomalies are shown in Table 2.
Figure 4 is a schematic diagram of six typical landslide relics in Nyingchi, created by superimposing optical satellite images with three-dimensional topographic features.
In Fig. 4(a), a lake has developed in the gully, with apparent topographic anomalies (chair-shaped terrain) and accumulation anomalies adjacent to the lake. Figure 4(b) shows topographic anomalies: the rear part of the landslide has a steep backwall, the middle part has a flat platform, and the front edge has small gullies. Figure 4(c) exhibits hydrological and vegetation anomalies, specifically river diversion, and the landslide shows less vegetation development than the surrounding environment. Figure 4(d) also shows the phenomenon of river diversion, along with a steep backwall and a fan-shaped accumulation, on which flow stripes have developed. The landslide relic in Fig. 4(e) exhibits typical texture anomalies, as evidenced by the rough image of the landslide body, which contrasts with the surrounding environment and displays color differences. Figure 4(f) exhibits topographic and hydrological anomalies, as well as signs of human activities, such as houses and cultivated land, on the accumulation body, suggesting that the landslide event occurred over a relatively long period.
In general, accurately identifying landslide relics requires a comprehensive analysis of multiple abnormal features, including terrain, water systems, and textures. When clouds appear in satellite images, it is also necessary to compare multiple image types for auxiliary interpretation. The Google Earth platform not only provides multiple types of optical and historical images, but also combines remote sensing imagery with terrain to create a realistic 3D model, thereby greatly improving the accuracy and efficiency of landslide interpretation.
4 Result
4.1 Characteristics of typical landslide relics
The study area contains numerous well-documented landslide relics, which not only supply fundamental data for this study but also help validate the inventory. As listed in Table 3, five previously reported landslides have all been identified in this work and are included in the inventory.
Jiabunong landslide (inventory No. 5130) is located in eastern Bayi, with a total vertical relief of 1492 m. It exhibits a chair-shaped rear scarp and a tongue-shaped frontal lobe, with an estimated volume exceeding 1 billion m
3 (
Du et al., 2021). The 102 landslide (inventory No. 61) lies south-west of Bomi, adjacent to the Purlung Tsangpo River. Previous studies have identified a cluster of landslide relics in this area, the largest of which corresponds to this landslide (
Wang et al., 2022b). Geduicun landslide (inventory No. 10376) is situated in eastern Zayu, near the Nu River. Its source area covers approximately 0.294 km
2, and the depositional area spans about 1.228 km
2 (
Zhao et al., 2022b). Signs of human activity, such as houses and farmland, are visible on the landslide mass, suggesting it occurred in the distant past. The Sedongpu landslide (inventory No. 5082) is located north-east of Mainling, near the Yarlung Zangbo River. This area has experienced multiple ice-rock avalanches historically, influenced by ice and snow melt (
Zhao et al., 2022a). The source zone features a typical chair-shaped back scarp, with small scraped gullies at the front. Yigong landslide (inventory No. 1043) occurred in 2000 in north-eastern Bayi. With an estimated volume of 3 Gm
3, it blocked the Yigong Zangbo River for 62 days; subsequent dam breach caused flooding along the Yarlung Zangbo River (
Shang et al., 2003;
Delaney and Evans, 2015).
4.2 Landslide relic inventory
Through this work, a total of 12,461 landslide relics were identified in the study area (Fig. 5(a)), covering 5179.059 km2. Among them, the largest landslide covers 20.910 km2, the smallest 0.499 km2, and the average 0.459 km2. The density of landslide relics in the study area is 0.083 per km2, and the landslide area density is 3.806%. There are significant differences in the distribution of landslide relics across different regions (Figs. 5(b) and 5(c)).
Statistical analysis of the number of landslide relics with different areas reveals a negative correlation between the area of the landslide relics and their corresponding numbers (Fig. 6). In the study area, the most significant number of landslide relics has an area of less than 1 km2, totaling 11207, accounting for 89.94% of all landslide relics. There are 915 landslide relics with an area of 1−2 km2, 206 with an area of 2−3 km2, and 133 with an area of more than 3 km2.
A detailed classification of landslide relics with an area of less than 1 km2 shows that over 65% have an area of less than 0.4 km2. Among them, the number of landslide relics with an area of 0−0.2 km2 is the largest, totaling 4977, accounting for 39.94% of the total number of landslides. The area of landslide relics between 0.2 and 0.4 km2 is the largest, totaling 926.364 km2 and accounting for 16.20% of the total area of landslide relics.
The spatial distribution characteristics of landslide relics can be characterized by point density and area density. In this paper, the kernel density calculation method is employed, with a search radius of 10 km. The calculated spatial distribution of landslide point density is shown in Fig. 7. It can be observed that there are four high-value areas of landslide point density in the entire Nyingchi area: the south-west of Nang County, the north-east of Mainling County, the south of Medog County, and the east of Zayu County. The landslide point density in these areas reaches 0.6 km−2, with a maximum of 0.9 km−2 (Fig. 7), located in the west of Nang County. The landslide point density in most other areas is below 0.5 km−2.
The spatial distribution of the area density of landslide relics is shown in Fig. 8. The area density of landslide relics in most regions is below 20%. The maximum landslide area density is 49%, located in the eastern part of Nang County (Fig. 8). The southern part of Nang County, the south-eastern part of Mainling County, and the eastern part of Zayu County are three high-density areas, all of which exceed 30%.
4.3 Correlation between landslide relics and environmental factors
Classification was performed based on the numerical characteristics of 9 environmental factors. Among them, continuous data such as elevation, slope, distance to rivers, distance to active faults, annual average temperature, and annual average rainfall were classified using the interval classification method. Discrete data, such as aspect, land type, and stratum, were classified according to their specific data values. Parameters such as the area of each sub-region, the number of landslides in the sub-region, and the area of landslides in the sub-region were counted, and the following formulas were used to calculate the landslide point density (LND) and area density (LAP) of each sub-region:
where N is the number of landslides in the sub-region, CA is the area of the sub-region, and LA is the area of landslides in the sub-region.
The area of each environmental factor zone and the corresponding distribution characteristics of landslide relics are shown in Figs. 9−11. It can be seen from Fig. 9(a) that as the elevation gradually increases, the area proportion of landslide relics increases accordingly, reaching the maximum between 3500 and 4000 m (with LAP being 5.98%), and then LAP gradually decreases. However, LND fluctuates with increasing elevation, with peaks appearing at intervals below 500 m, 2000−2500 m, and 4500−5000 m. Moreover, the number of landslide relics below 500 m elevation in the study area is the largest, with LND reaching 0.13 km−2; however, the proportion of landslide area in this interval is relatively small, indicating that small-scale landslides are primarily developed at elevations below 500 m.
For slope, the LAP and LND of landslide relics were calculated in 20° intervals. Both LAP and LND peak within the 20°–30° slope range, reaching 5.28% and 0.11 km−2, respectively (Fig. 9(b)).
Aspect distribution in the study area is relatively uniform. The highest LND value occurs on SSW (0.09 km−2), while the highest LAP value is observed on SWW (4.68%) (Fig. 9(c)).
The study area comprises seven land use types, ranked in descending order of area coverage: forest, grassland, snow/ice, barren, water, cropland, and shrub. Shrublands exhibit the highest number of landslide relics (0.39 km−2), followed by croplands (0.14 km−2) and grasslands (0.13 km−2). In terms of landslide scale, croplands have the largest LAP (12.08%), followed by shrublands (7.04%) and grasslands (5.83%) (Fig. 9(d)).
The study area comprises 12 stratigraphic types: Quaternary, Neogene, Paleogene, Cretaceous, Jurassic, Triassic, Permian, Carboniferous, Devonian, Ordovician, Lower Paleozoic, and intrusive rocks. Among these, intrusive rocks constitute the largest areal proportion at 34%. However, the Triassic strata exhibit the highest concentration of landslide relics in terms of both number and area, with an LND of 0.23 km−2 and an LAP of 11.75%, significantly exceeding those of other strata (Fig. 10(a)).
Multiple active faults are present in the region. The distribution of landslide relics was analyzed in 10-km intervals from these faults. Overall, LAP generally decreases with increasing distance from active faults, while LND exhibits fluctuating trends. The maximum LAP (4.55%) occurs 20–40 km from faults, whereas the highest LND (0.10 km−2) is observed 120–140 km away. Both LAP and LND reach their minimum values beyond 140 km, at 1.10% and 0.04 km−2, respectively (Fig. 10(b)).
Analysis of annual average temperature from 1901 to 2023 shows that most of the study area experiences temperatures between –5°C and 0°C (32.9% of the total area), followed by the 0°C–5°C (19.3%). The highest density of landslide relics (0.12 km−2) occurs in regions with temperatures above 20°C. The –5°C–0°C and 0°C–5°C ranges rank second and third in LND, at 0.11 km−2 and 0.09 km−2, respectively. However, the largest proportion of the landslide area (5.56%) is found in the 0°C–5°C zone (Fig. 11(a)).
Spatial patterns of mean annual precipitation align with those of temperature. The highest LND (0.29 km−2) and LAP (16.4%) both occur in areas with precipitation below 400 mm. The 2200–2400 mm precipitation range ranks second in LND (0.20 km−2), but accounts for a relatively small landslide area (2.01%) (Fig. 11(b)).
The distance between landslide relics and rivers was statistically analyzed. The highest LND (0.09 km−2) is found within 1 km of rivers, while the largest LAP (4.12%) occurs 1–2 km away. Both the number and area of landslide relics generally decrease with increasing distance from rivers (Fig. 11(c)).
Landslide distribution relative to roads was assessed in 20-km intervals. Both LAP and LND peak within 0–20 km of roads, at 3.60% and 0.08 km−2, respectively. LAP decreases monotonically with distance. LND declines up to 120 km, then increases, reaching a secondary peak of 0.66 km−2 in the 160–180 km interval (Fig. 11(d)).
5 Discuss
5.1 Differences from the existing landslide inventory in the study area
We compiled 10 landslide databases for the Nyingchi area and its vicinity, with specific details provided in Table 4. The landslide inventory presented in this paper comprises 12461 landslides covering a total area of 5179.059 km2, which is significantly larger than the published landslide inventories in the Nyingchi area and its surrounding areas. This indicates the meticulousness and comprehensiveness of this inventory’s construction.
In terms of the scope of landslide inventories, it can be observed that those in the Nyingchi area and surrounding regions primarily focus on three themes: along major projects, key parts of geological structures, and administrative divisions. Specifically, No. 3, No. 4, No. 5, No. 7, No. 8, and No. 9 are landslide inventories along the Sichuan‒Xizang traffic corridor (STTC), while No. 10 is the landslide inventory along the China-Nepal transportation corridor; the databases of No. 2 and No. 6 are landslide inventories constructed against the background of key tectonically active parts of the Qinghai-Xizang Plateau; No. 1 and No. 10 are landslide inventories constructed with administrative divisions as the geographical scope. Finally, landslide database construction primarily uses four methods: remote sensing interpretation, literature review, field investigation, and InSAR analysis. Among them, InSAR analysis requires comparing SAR time series, and the resulting landslide data are those with recent deformation. However, this method does not account for potential landslides, which may lead to missing data in the landslide database and thus affect the results of the landslide susceptibility assessment.
Based on the above understanding, it can be confirmed that the database constructed in this study takes administrative divisions as its geographical scope and adopts the currently mature methods for building landslide inventories. The constructed landslide relic inventory is the largest in terms of quantity and area of landslide relics in the south-eastern Qinghai-Xizang Plateau to date.
5.2 Controlling effect of environmental factors on the spatial distribution of landslide relics
5.2.1 Topographic and geomorphic factors
Topographic and geomorphic factors are the primary conditions for landslide occurrence. Among the environmental factors selected in this study, the topographic and geomorphic factors include elevation, slope, aspect, and land use types. With increasing elevation, the degree of development of landslide relics shows two peaks at the 1500−2500 m and 3500−5000 m intervals (Fig. 9(a)). Two factors mainly control this. First, the 3500−5000 m elevation interval is the largest area in the study area, and the degree of development of landslide relics reaches a peak in this interval. Secondly, the 1500−2500 m elevation area is mainly located in the southern part of Nyingchi, where the overall landslide development is relatively low. However, in Manigang Village, located in the western part of Medog County, numerous landslide relics have developed along both sides of the river channel. The area density of these relics in this area exceeds 20% (Fig. 8). Satellite imagery reveals the presence of a soil-rock road in Manigang Village. Notably, the construction and expansion of such roads have been identified as a key factor contributing to the substantial increase in both the frequency and scale of landslides in the Himalayan region (
Saha and Bera, 2025). Drawing on this finding, it is reasonable to hypothesize that engineering construction activities in the western part of Medog County have disrupted the regional stress balance. This disruption, in turn, has facilitated the development of numerous landslide disasters along the banks of rivers and the sides of highways in the area.
In terms of aspect, the number of landslides in the SSW and SWW directions in the study area is slightly more than that in other directions (Fig. 9(c)). At the same time, the degree of development of landslide relics in these two aspects is also slightly higher than in other aspects. This indicates that tectonic activity controls the distribution of landslide relics (
Görüm, 2019;
Obda et al., 2024). Because the Nyingchi area is strongly affected by the nearly north–south compression of the Himalayan orogenic belt, more slopes in the SSW and SWW directions are developed, leading to more landslide events in these directions.
In terms of land types, the development degree of landslides on shrubland is the highest, followed by cropland (Fig. 9(d)). However, the area of shrubs is small and distributed too scatteredly. Only 71 landslides have developed on the shrubland, covering a total area of 12.65 km
2. This result cannot indicate that the shrub is conducive to the development of landslides. The total area of cropland in the study area is also relatively small, but its distribution is concentrated. For the convenience of planting, cropland is mainly distributed along river channels, in valleys, and in residential areas, which are also high-incidence areas for landslides (
Rahim et al., 2021). Although the forest area in the study area is large, the number and scale of landslide relics are smaller than those in other land types, indicating that tall arbors in Nyingchi can effectively reduce the occurrence of landslide disasters.
5.2.2 Geological structural factors
Geological structural factors include two components: stratum and distance to active faults. Although the Triassic ranks fourth in Nyingchi in terms of area, the degree of development of landslide relics in the Triassic is significantly higher than that in other strata. This characteristic is similar to the development characteristics of landslide relics in the Lhasa area (
Ma et al., 2025). After analysis, it is believed that two main reasons control this. First, the Triassic in the study area is primarily distributed on the southern side of the Yarlung Zangbo fault and on both sides of the Nujiang fault, providing external dynamic conditions for landslide development. Secondly, the Triassic lithology is primarily composed of easily weathered rocks, such as sandstone, tuff, and slate, which provide favorable conditions for landslide development.
Regarding the distance to active faults, LAP shows a gradual decrease with increasing distance, whereas LND shows fluctuations. It can be inferred that fault activity is only one external factor controlling the development of landslides in the Nyingchi area, and other factors, such as rainfall and human activities, may also influence landslides in this area. In the 100−120 km and 120−140 km intervals, the number of landslides increases significantly. These areas are all located in the western part of Gongbo’gyamda County, with the Carboniferous as the main stratigraphy. Although active faults are less developed, the Carboniferous is primarily composed of terrigenous clastic rocks, such as silty slate and quartz sandstone, which provide favorable geological conditions for landslide development, resulting in a significant increase in the number of landslides in this area.
5.2.3 hydrometeorological factors
Based on the hydrometeorological characteristics of the Nyingchi area, we selected the distance to rivers, annual average precipitation, and annual average temperature to analyze the relationship between these factors and the degree of development of landslide relics. Regarding the distance from rivers, the degree of landslide development decreases with increasing distance from rivers, consistent with the general law of landslide development (
Shao et al., 2020;
Wang et al., 2024).
The spatial variation in annual average precipitation in the Nyingchi area is significant, leading to a complex relationship between the development characteristics of landslide relics and rainfall. The area with the lowest annual average precipitation is around Nang County in the western part of the study area; however, the landslide development in this area is relatively high. Therefore, the development of landslides in this area is mainly controlled by tectonic activities. With increased rainfall, the degree of development of landslide relics remains relatively stable. However, in areas with rainfall above 2000 mm, the number of landslide relics increases significantly, but the landslide area decreases instead. These areas are concentrated in the southern part of Medog County. The region features low altitude, gentle terrain, predominantly forested landscapes, and high annual precipitation (
Wang et al., 2022a;
Dou et al., 2023), characterized by a typical subtropical monsoon climate (
Wang et al., 2023). Therefore, it is speculated that rainfall-induced group landslides mainly develop in the southern part of Medog County. This type of landslide is characterized by a large number, small volume, and shallow sliding surface (
Van Asch et al., 1999;
Marc et al., 2018;
Mao et al., 2024), resulting in an increase in LND value and the maintenance of a low LAP value in the area with an annual average precipitation above 2000 mm in the study area.
With increasing annual average temperature, LAP shows a first-increase-then-decrease trend, and LND broadly follows that of LAP. However, in areas with temperatures above 20°C, the LND value suddenly increases and reaches its maximum. It can be observed that, except for the area with temperatures above 20°C, the highest LAP and LND values are within the range of −5°C to 5°C. The freeze–thaw action in this temperature range is the strongest, which provides relatively favorable conditions for the development of landslides (
Hu et al., 2019;
Huang et al., 2025). The area with an annual average temperature above 20°C in the study area highly overlaps with the area with annual average rainfall above 2000 mm. Therefore, the frequent occurrence of rainfall-induced group landslides in the southern part of Medog County is the primary factor contributing to the increase in LND value and the maintenance of low LAP value in the area, where the annual average temperature exceeds 20°C.
5.2.4 Human activity factors
The population in the research area is sparse, and the residential areas are relatively scattered. The human activity that significantly impacts the stability of the rock mass is road construction (
Mao et al., 2024). Therefore, this paper chooses the distance from the road to represent human activity factors. Overall, human activities promote landslides; that is, LAP monotonically decreases with increasing distance from the road. Between 0 and 120 km, LND shows a decreasing trend, then gradually increases, reaching a peak in the 160−180 km interval. The area, located 160−180 km away from roads in the study area, is mainly situated in the southern part of Medog County, where human activities are relatively scarce. However, the river system here is well-developed, with the Xibaxia Qu, which has the third-largest water volume in Xizang, after the Yarlung Zangbo River and the Nujiang River. As inferred earlier, this area is controlled by hydrometeorological factors and is dominated by rainfall-induced group landslides. These factors, together, lead to an increase in LND in the region, 160−180 km away from roads.
Analysis shows that geological structure and topography, as the core dominant factors, jointly determine the spatial pattern of high-value landslide areas and are the first controlling factors for regional landslide development. Specifically, this is manifested in the formation of four high-risk areas in the south-west of Nang County, the north-east of Mainling County, the south of Medog County, and the east of Zayu County. Hydrometeorology and human activities are auxiliary factors, and their scope of action is limited to high-value areas that are controlled by dominant factors. They only change the number or scale of landslides through triggering and amplification effects, forming the second controlling factor for regional landslide development. Furthermore, there are regional differences in the intensity of the role of auxiliary factors: in the western regions where geological structures and terrain have significant effects (Nang County, Mainling County), the influence of auxiliary factors represented by river lateral erosion is more prominent; in the eastern region (southern Medog County) where geological structures and terrain have weak effects but strong rainfall conditions, hydrometeorological auxiliary factors such as heavy rainfall have become key variables in regulating the number of landslides. From the above, it can be seen that the spatial distribution of landslide relics in Nyingchi is the result of the comprehensive effects of four factors: topography, geological structure, hydro-meteorology, and human activities. It is challenging to establish a clear, linear mathematical relationship between a single control factor and the degree of development of landslide relics. Therefore, artificial intelligence methods based on big data analysis and machine learning are the most optimal technical approach for conducting landslide susceptibility assessments in the region. The inventory of landslide relics presented in this paper will provide sufficient and accurate basic calculation data to support the construction of landslide susceptibility assessment models in Nyingchi in the future.
6 Limitations and prospects
6.1 Study limitations
This study systematically constructed a landslide relic inventory for Nyingchi, defined by the administrative boundaries of Nyingchi. A comparative analysis with existing research catalogs reveals that this inventory achieves significant breakthroughs in terms of quantity, coverage area, and spatial integrity of landslide relics—particularly by filling the survey gap in areas beyond major transportation routes that existed in previous studies. However, restricted by limitations in research methods and the quality of remote sensing images, this inventory has not yet fully included all landslide relics in Nyingchi, with specific issues in the following two aspects.
First, the human-computer interactive visual interpretation of large-scale landslide relics requires collaboration among multiple technicians. However, subjective differences exist among technicians regarding interpretation standards for landslide relics, which may lead to deviations in interpretation results and affect data accuracy. Second, due to the complex natural conditions in the study area, remote sensing images cannot clearly present surface features in some regions, particularly in areas with snow and ice coverage and cloud obstruction, which significantly increases the difficulty of identifying small-scale landslide relics. This problem is more prominent in high-altitude areas above 5000 m above sea level in the northern part of the study area, where snow and ice coverage is extensive.
To address the above limitations, future studies will carry out optimization work from three aspects: first, combining UAV aerial photography technology with field surveys to conduct supplementary investigations in potentially missed areas, further improving the inventory data; second, establishing a cross-validation mechanism for the landslide relic inventory, reducing errors caused by inconsistent interpretation standards and avoiding major interpretation mistakes through mutual verification of interpretation results by multiple technicians; third, expanding the historical remote sensing image data source, selecting images with complete time series and high clarity for re-interpretation, and enhancing the support capability of image data for landslide relics. In the future, regular UAV aerial photography and field surveys will also be conducted in key areas of Nyingchi to dynamically verify and update the landslide relic inventory, ensuring the integrity and accuracy of the inventory to the greatest extent.
6.2 Application prospects
The landslide relic inventory developed in this study for Nyingchi has clear practical application value and can serve as a supporting role in three major fields: engineering construction, geological disaster prevention and control, and territorial spatial planning. The specific application directions are as follows.
6.2.1 Engineering site selection and risk avoidance
This inventory can serve as a basic geological data source for major projects, such as the Nyingchi-Zhayu section of the Sichuan-Xizang Railway and hydropower projects in the lower reaches of the Yarlung Zangbo River. By conducting spatial overlay analysis between high-density landslide relic areas and engineering routes or project site selection scopes, potential landslide risk sections in engineering construction can be accurately identified. This provides a scientific basis for optimizing and adjusting engineering routes, as well as for selecting and designing protective projects, thereby reducing safety risks and economic losses associated with landslide disasters in engineering construction.
6.2.2 Geological disaster monitoring and early warning
Based on the spatial distribution pattern of landslide relics in the inventory, key monitoring areas for landslide disasters in Nyingchi can be delineated. Combined with InSAR deformation data, the risk of landslide reactivation is predicted by analyzing current deformation trends in historical landslide areas. Meanwhile, potential hidden dangers of new landslides are identified according to abnormal deformation signals in non-landslide areas. This not only enhances the accuracy and timeliness of regional landslide disaster early warnings but also provides dynamic data support for informed disaster prevention and mitigation decisions.
6.2.3 Optimization of territorial spatial planning
The correlation between landslide relic distribution and land-use types is integrated into Nyingchi’s territorial spatial planning system. In urban expansion planning, low landslide risk areas are prioritized as development spaces, and large-scale concentrated construction projects, such as residential and commercial projects, are avoided in high landslide risk areas. In the optimization of agricultural layout, crop cultivation and animal husbandry are guided to concentrate in gentle areas with low landslide risk, reducing damage to agricultural production caused by landslide disasters. At the same time, high landslide risk areas are included in ecological protection red lines or designated as geological disaster prevention and control areas, thereby realizing the coordinated advancement of territorial spatial development and geological disaster prevention and control.
7 Conclusions
1) Based on administrative boundaries, this study delineated the research area and established an inventory of landslide relics in the Nyingchi region, effectively supplementing landslide data for areas lacking residents and infrastructure within the study area and filling the prior data gaps.
2) The Nyingchi landslide relic inventory comprises 12461 entries, with a total coverage area of 5179.059 km2 and an average of 0.459 km2 per landslide. Individual landslide areas range from 499 m2 (minimum) to 20.910 km2 (maximum). To date, this inventory represents the most comprehensive regional landslide data set, outperforming all existing counterparts in both landslide count and total coverage.
3) Landslide relic point density exceeds 0.6 km−2 in south-western Nang County, north-eastern Mainling County, southern Medog County, and eastern Zayu County, defining these as high point-density areas. Landslide area density exhibits a spatial pattern broadly consistent with point density: exceeding 30% in southern Nang County, north-eastern Mainling County, and eastern Zayu County, which constitute the three primary area-density high-value zones.
4) The spatial distribution of landslide relics in Nyingchi is jointly influenced by four factor categories: topography and geomorphology, geological structures, hydrometeorological conditions, and human activities. Among these, geological structures and geomorphology are the dominant core factors, jointly shaping the spatial pattern of high-potential landslide zones and serving as the primary controls on regional landslide occurrence and development. Hydrometeorological conditions and human activities, by contrast, function as auxiliary factors—their influence constrained to the dominant factor-delineated high-potential zones, merely modifying landslide frequency, magnitude, or scale via triggering and amplifying effects, and thus serving as secondary controls on regional landslide dynamics.
PDF
(13969KB)
