Risk assessment of geological disasters in Nyingchi, Tibet

2.1 Overview of the studied area

Nyingchi is located in the southeast of Tibet Autonomous Region, with geographic coordinates of 92.165°–98.753°N and 27.559°–30.669°E (see Figure 1). This area is located at the junction of the Eurasian and Indian Ocean plates, with fractured strata and active crust. Several fault zones are densely distributed here, including (1) Jiali-Chayu fault zone, (2) fault zones of the main boundary and the main front thrust of the southern Himalayas, (3) Yarlung Zangbo fault zone, and (4) their secondary fault zones [8] (see Figure 2). The average altitude is around 4,000 m. However, the highest altitude can reach 7229.72 m, whereas the lowest altitude is only 151.55 m. The world’s deepest canyon, the Yarlung Zangbo River Grand Canyon, is also located here, making Nyingchi a region with relatively large vertical drop of landform on land [9]. Large surface undulations and remarkable topographical change create coexistence of multiple climates in the region. The special tropical humid and semi-humid climates make the vegetation coverage rate as high as 46%. The water system formed by the Yarlung Zangbo River and its tributaries causes abundant precipitation but with uneven distribution. These special geographical environments have become an important reason for the frequent occurrence of geological disasters in this area.

Figure 1

Map of studied areas in Nyingchi.

Figure 2

Map of Nyingchi fault zone.

2.2 Types and distribution characteristics of geological disasters in Nyingchi

Frequent geological disasters in Nyingchi area are mainly landslides, collapses, mudslides, frozen soil slopes, and glacier melting. There are about 500 recurrences in recent decades. They are mainly distributed in the upper and middle reaches of the Yarlung Zangbo River and the Himalayan fault zone. The impact of earthquakes and heavy rains causes the occurrence of one type of geological disasters to trigger others and secondary disasters. The ranges of occurrence are wide, and their frequencies are high with complex internal and external coupling mechanisms. The resultant harm is significant. Together with various human engineering activities, these factors create massive difficulty to restoration and seriously limit the construction of infrastructures such as highways, power grids, and telecommunications in the area. Figure 3 shows a landslide disaster in a section of Highway 318.

Figure 3

Typical site landslide map.

Landslides and mudslides are mainly affected by the melting of glaciers and mountain snow or heavy rainfall within a short period of time. Earthquakes, fault zone activities, and blasting in engineering construction may cause collapse, which usually forms barrier lakes and induces mountain torrents. Casualty and property loss are significant. According to statistics, there were more than 110 earthquakes with magnitudes beyond 4 from 1980 to 2018. The quake in Milin in 2017 had magnitude 6.9, causing a large-scale collapse and a barrier lake [10]. The 8.6 magnitude earthquake of Chayu in 1950 caused a composite geological disaster consisting of collapse, landslide, and debris flow, blocking the road for 1 year [11]. In 1959, the giant mountain in Basu collapsed and interrupted the river, forming Lake Ranwucuo. The enormous landslide in Layue in 1967 was mainly affected by the activity of the fault zone. Two years before the landslide, there were seven earthquakes in this area with magnitude 4 or higher. Some of the severe landslides caused by melting of glaciers are as follows: (1) Guxiang landslide in 1953, (2) Palongzangbo 102 landslide in 1982, (3) group landslides in Dongjiu in 1991, and (4) Yigong landslide in 2000. There are more than 2,000 landslides with various scales in recent 60 years, bringing continuous damage to roads [12]. The geological structure of Pailonggou mountain is loose. At least 50 geological disasters, which are mostly mudslides, occur every year during the rainy season from July to September. They directly destroy the only road, block the traffic, and cause long-term power outage. The Chayu County severely suffers from mountain floods with 62 occurrences in 15 years. In both 2007 and 2018, giant debris flows occurred in Tianmogou resulting in burst of the dammed lake, which flooded downstream villages and caused significant casualties [13].

3.1 Data sources

For the basic data sources for the geological disaster risk assessment factors used in this study, see Table 1.

Figure 4

AHP model.

3.2 Methodology

3.2.2 Fuzzy mathematics

3.2.2.1 Establishment of evaluation indicators

According to the different risks of natural disasters in the study area, the selected nine sub-criteria-level impact factors use the expert system to classify the indicators at two levels: basic factors (μ ₁) and inducting factors(μ ₂); and build a risk evaluation index system [17].

3.2.2.2 Establishment of evaluation sets

The risk level of the study area is divided into five levels, that is, V = (V ₁, V ₂, V ₃, V ₄, V ₅) = (I, II, III, IV, V) = (very low, low, medium, high, very high) (see Table 2).

Table 2

Evaluation set

Evaluation factors	Danger classes
	Very low	Low	Medium	High	Very high
Slope	0–16.48	16.48–28.04	28.04–38.56	38.56–51.53	51.53–89.37
Classification of land use	Mid-high plain	High plain	Low, mid, mid-high mountain	High, extreme high mountain	Glacier
NDVI	0.83–0.99	0.65–0.83	0.44–0.65	0.21–0.44	<0.21
Distance to flaunts	>4,000	3,000–4,000	2,000–3,000	1,000–2,000	0–1,000
Distance to rivers	>2,000	1,500–2,000	1,000–1,500	500–1,000	0–500
Annual precipitation	428–558.8	558.8–689.6	689.6–820.4	820.4–951.2	951.2–1,082
Landform	Build-up land	Arable land	Woodland, Grassland	Water	Badlands
Relative elevation	0–56	56–180	180–527	527–1,239	1,239–5,830
Aspect	Flat	South	Southeast, southwest	East, west, northeast, northwest	North

3.2.2.3 Establishment of membership function

The correct construction of the membership function is a very critical step in single-factor evaluation in fuzzy comprehensive evaluation, that is, appropriate parameters are usually selected for fitting according to the actual value of the required conclusion. In practice, the parameter forms often used include triangle, normal, and trapezoid. In this paper, the quantitative index chooses the trapezoidal fuzzy number that has been used in extensive research and the expression is as follows:

(5) μ A ( x ) = 0 x < a x − a b − a a ≤ x < b 1 b ≤ x ≤ c d − x d − c c < x ≤ d 0 x ≥ d ,

where μ _A(x) is the membership function of each evaluation index, x is the measured value of the evaluation index, a ₁ = V ₁, a ₂ = (V ₁ + V ₂)/2, a ₃ = (V ₂ + V ₃)/2, a ₄ = V ₃, V ₁, V ₂, V ₃ are the grading thresholds of the evaluation index corresponding to the risk level of collapse from low to high. The schematic diagram of the membership function is shown in Figure 5.

Figure 5

Schematic diagram of membership function.

4.1 Selection process

Occurrence of geological disasters is usually caused by the coupling of basic and inducing factors. Combining Nyingchi’s natural conditions and human factors, while considering the size of the studied area, availability of evaluation factors, and requirements of the research accuracy, nine evaluation factors are selected including landform, hillside slope, hillside aspect, relative elevation, distance to rivers, annual precipitation, distance to fault zones, land use type, and normalized difference vegetation index (NDVI). Using 30 × 30 m grid units, the model of AHP-Geological hazard risk assessment is established. Three software programs, such as ArcGIS10.2, Envi5.1, and MRT, are used for in-depth processing of the basic data. The operation process is illustrated as follows (see Figure 6).

Figure 6

Flowchart of factor extraction.

Figure 7

Geological map.

Figure 8

(a) Landform. (b) Grade. (c) Aspect. (d) Relative elevation. (e) Distance to rivers. (f) Annual precipitation. (g) Distance to faults. (h) Classification of land use. (i) NDVI.

4.2 Selection results

5.1 Evaluation results

Nine evaluation factors are assessed using the analytic hierarchy process. The judgment matrix of the factors is calculated, and its consistency test is performed. The weight distribution W _i , the maximum eigenvalue λ _max, and the consistency ratio CR are derived (see Table 3). The CR being less than 0.1 demonstrates that the judgment matrix has good consistency and meets the consistency test standard. According to equation (4) combined with the contribution value of evaluation factors to each scheme, the GDHI is calculated (see Table 4).

1. The sensitivity and evaluation results of the two methods to the risk area are obviously different. After AHP only uses the expert scoring method, it obviously widens the gap between the five evaluation grades, while FAHP combines quantitative and qualitative methods to make the evaluation results more in line with the actual situation. The extremely low- and extremely high-risk areas are only a few.

2. The slope accounted for 32.62%, which is equivalent to the distance from the fault zone and the relative height difference of the topography but much higher than the other six items. This is because the slope in this area is steeper than other areas in Tibet. Mainly concentrated at 28–51°, which is the geographical basis for natural disasters that cannot be ignored; the distance from the fault zone is slightly higher than the weight of land use type, NDVI, distance from water system, and landform-induced geological disasters. The average annual precipitation and slope influence are far less than that of the fault zone, accounting for 3.5% and 2.36% respectively.

3. Large-scale rainfall in a short period of time has played an important role in landslides and mudslides. However, as the average annual rainfall in this article is only 3.5%, according to the expert scoring method and the analysis of the actual disaster situation, to avoid the accumulation of errors, the weight of this factor was actively reduced. More than 80% of the huge landslides, mudslides, and collapse disasters that have occurred in Nyingchi have occurred after continuous rainfall in summer or after the melting of glaciers caused by temperature rise, and very few of them are caused by human engineering activities.

4. The four types of land use, NDVI, distance from water system, and landform have a strong correlation and great mutual influence, and their impact on geological disasters is basically the same. The proportions are 6.42, 7.17, 6.42, and 7.35%, respectively, showing a uniform distribution.

5. For different levels of risk areas, the contribution values of various evaluation factors are also different. Among the extremely low-risk areas, only the distance from the fault zone contributed the most prominently, reaching 47.06%; in the low-risk area, the contribution of vegetation coverage, distance from the water system, topography, and relative height difference are similar, exceeding 35%; in the medium-risk zone, slope and annual average precipitation accounted for the highest proportions of all types, 47.72 and 47.95%, respectively, but the vegetation coverage and distance from the water system only accounted for 4.94 and 4.76%; land use types, NDVI, and distance from water systems in high-risk areas contribute about 30%; the aspect ratio of the extremely high-risk area is 42.75%, ranking first, and the remaining factors contribute relatively evenly.

5.2 Risk assessment

In the ArcGIS software, GDHI is weighted and superimposed to obtain the Nyingchi geological hazard risk assessment map (see Figure 9). The Nyingchi area is divided into five grades using the natural discontinuity method. The derived evaluation chart is further analyzed with Figures 1 and 3 being the base map (see Table 4).

Figure 9

Geological hazard risk assessment map of Nyingchi.

The extremely low-risk area is 16,330 km², accounting for 15.34% of the total area. It is mainly distributed in the central part of Sijol River, the central part of Yarlung Zangbo River, and the marginal area of Chayu County; the low-risk area is 30,130 km², accounting for 28.31% of the total area. It is mainly distributed along the river systems of Yigong Zangbu, Palong Zangbu, Yarlung Zangbu, Danba Qu, Sangqu, Nu River, Chayu Qu, Gongri Gabu Qu, Siba Xia Qu, and Sijol River, especially in the main rivers of Metuo County, Mirin County, and Chayu County; the middle-risk area is 29,430 km², accounting for 27.62% of the total area. Some of them are distributed along the upper and lower reaches of the Brahmaputra, the upper and lower reaches of the Yarlung Zangbo River, the Siba Xiqu, the Nu River, and Chayu Qu. The rest are mainly in the northwest of Paki District, the middle part of Mirin County, most of Motuo County, and the west and southeast of Chayu County; the high-risk area is 18,430 km², accounting for 17.33% of the total area, showing a decreasing trend from north to south. It is evenly distributed across Gongbujiangda County and the middle-risk area. As the fault zone and the junction of counties are basically coincident, these are also in the high-risk area; the extremely high-risk area covers an area of 12,430 km², accounting for 11.65% of the total area. It is mainly distributed in Bhumi County, in the north of Chayu County, and in the southeast of Motuo County. The three counties also present extremely high risk at the intersection.

5.3 Accuracy verification

A point in the low-risk area of the obtained hazard assessment map is randomly selected. The assessment results of selection are verified through historical images of Google Earth in 1984, 2004, and 2013 and the actual field survey in July 2020 (see Figure 10). In 1984, the site suffered geological disasters all the year round. It was in the initial period of geological disasters, and the scale was above the middle level. As there was no human inhabitation, no preventive actions were implemented. In 2004, ecological restoration was carried out by planting trees, aiming for reducing disaster occurrence. However, the effect was not obvious as the mountain is steep, the vegetation is sparse, and the terrain has a relatively large elevation difference (more than 500 m). The geological disasters that occurred later were still severely destructive. In 2013, the project was improved by strengthening and constructing a drainage system for further control. Residential land was also established on the opposite bank of the river. When visiting the site in the rainy season of 2020, there was only slight landslides and mudslides with a small amount of rubble rolling down. Threats to the life of nearby residents and construction activities are insignificant.

Figure 10

Verification point diagram.