1 Introduction

Many earthquakes have occurred in China in the 21th century. In the last 5 years, two high-magnitude earthquakes occurred in Sichuan Province (the Ms 8.0 Wenchuan earthquake and the Ms 7.0 Lushan earthquake) [1–3]. Both earthquakes induced a large number of landslides that caused serious damage to roads. Statistical data show that landslide disasters triggered by the Ms 8.0 Wenchuan earthquake damaged nearly 28,000 km of highway (including 21 expressways, 15 national and provincial highways, and 2,795 rural highways) and caused damage to railways (including the Chengguan Railway, Guangyue Railway, Detian Railway, and the Mianyang–Guangyuan segment of the Baocheng Railway). The damaged roads not only caused huge losses of life and property, but also blocked traffic and seriously delayed the rescue operations. Maintaining smooth traffic flow was very challenging [4].

Engineers and transport experts should pay close attention to regional geology when selecting the routes for roads running through rugged mountainous areas based on practice [5]. For a full understanding of the geological disasters and geo-hazards in the route region, detailed geological exploration data must be obtained. However, during the feasibility study stage of route selection, the route schemes are yet to be determined; therefore, generally, no large-scale geological investigation is performed. Hence, it is important to obtain as much geological hazard information as possible in the route area to guide the route planners and engineers by using existing satellite remote sensing data.

Landslides cause serious damage in mountainous regions. Currently, analysis methods of regional slope stability under seismic action include the pseudo-static method and the Newmark slider analysis method. To estimate slope stability, the minimum coefficient of slope stability under seismic action was obtained by Leshchinsky [6], Ling [7], and Biondi [8], and by other methods based on the pseudo-static method. However, there are obvious deficiencies in the pseudo-static method. First, the seismic inertia force is neither invariable nor unidirectional, with rapid fluctuations in both magnitude and direction. Second, a slope coefficient of stability less than 1 will not necessarily lead to an overall slope instability, but may only create a permanent slope displacement. Newmark [9] calculated the deformation of a dam caused by an earthquake using the slider model. Subsequently, the Newmark method was combined with GIS by Jibson [10–12], Miles [13], and Rodríguez-Peces [14]. Using GIS technology, they divided the study area into grids of different sizes and each grid was seen as an independent slope. Then the seismic landslide hazard classification for the study area was performed according to the displacement size. Although this method has great advantages for prediction of regional landslide triggered by earthquakes, the stability of slopes is also strongly affected by the structural features and characteristics of the slope, the rock and soil composition, and the general slope of the terrain. Thus, taking these factors into consideration during landslide hazard assessment can greatly improve the accuracy of the predictions.

To improve the estimation accuracy, a method for rapid assessment of seismic landslide hazards was developed. This method uses GIS technology, and divides the slope units according to the geological and hydrogeological conditions. A hazard-zone map for landslides triggered by regional seismic activity can be generated rapidly based on a comprehensive index method. This method can be used to obtain guidelines for initial route selection in high seismic regions and for reconstruction projects along existing routes. This can significantly reduce the work required during the feasibility stage of route selection, and narrow the scope of the geological exploration required for project planning. Correct prevention and mitigation measures can be proposed, project safety improved, and losses caused by design changes reduced.

2 Principle and method of seismic landslide hazard zoning

The principle of the rapid assessment of seismic landslide hazard zones is based on satellite remote sensing data. Combining GIS spatial analysis techniques and statistical methods for seismic landslide analysis, the slope units are determined along pre-built routes and existing reconstruction routes. Then the statistics, identification, and analysis are performed on the slope units. Finally, the seismic landslide hazard zoning map can be generated. The block diagram of the identification method is shown in Fig. 1.

Fig. 1
figure 1

Block diagram of assessment of seismic landslide hazard zones

Full size image

2.1 Determining the units of the regional slopes

2.1.1 Division of slope units

At present, the slope units are divided mainly into two types: regular grid units and complex slope units. The regular grid units can be easily processed, using a relatively simple calculation method and algorithm. However, this evenly distributed grid unit is not related to the actual failure unit and physical mechanism. It is completely unrelated to the geology, topography, and hydrology. The slope unit is defined as an area with features that are distinct from those of the adjacent regions. The formation of the terrain reflects the long-term effects of the geological conditions and hydrogeological conditions [15–17]. The schematic diagram of the slope (Fig. 2) shows the ridge line and the valley line of the digital topographic map, found through the hydrologic analysis functions of ArcGIS. The catchment area was divided into slope A and slope B defined by the valley lines and ridge line.

Fig. 2
figure 2

Schematic diagram of the slope used in our method

Full size image

The size selection of the slope units depends on the average size of the historic landslide in the study area. For objects with similar topography, geology, and hydrological conditions in the same area, we assume a similar the scale of the past and future landslide failure slopes. When dividing the slope units, the threshold value of the catchment area determines the size of the slope unit. In practice, multiple threshold values are needed and trial calculations are performed to determine the appropriate threshold level.

2.1.2 Hydrological analysis

The hydrological characteristics of the study area are related to the topography of the region. During the rainy season in the mountains, some water infiltrates into the ground, and the remaining flows from higher to lower elevation under the force of gravity. Hydrological analysis of the area was carried out using the hydrologic analysis tools of ArcGIS. The hydrological analysis module was used to extract the depression-free digital elevation model (DEM), flow direction, flow collection, channel network generation date, and basin analysis from the DEM. Then the ridge line was obtained. The valley line was extracted by changing the original valley in the model to a ridge. The same method was used to extract the depression-free DEM, flow direction, flow collection, channel network generation, and basin analysis from the inverted DEM and obtain the valley line. When transforming the raster data into polygons, some incorrect polygon slope units may appear: very small polygon slope units and inconsistent polygon slope units. These incorrect polygon slope units can be eliminated by using the ‘dissolve’ function and the ‘integrate’ function in GIS. If the result of the division of the polygon slope units is still unsatisfactory, it can be adjusted by artificial modification.

2.2 Identifying slopes that threaten transport routes

The safety of railway and highway routes in the study area is at high risk from landslides, particularly if the axis of the landslide and the axis of the route intersect at a large angle. A landslide can cause serious damage and injuries when this angle ranges from 60° to 120° (an angle between the slope and the route alignment of −30° to +30°), as shown in Fig. 3.

Fig. 3
figure 3

Illustration of slope angles affecting landslide risk

Full size image

2.3 Prediction of seismic landslide risk based on the comprehensive index method

Based on the obtained map of unstable slope zones in the study area, the comprehensive index method [18] is used to predict seismic landslides that threaten the transport route. This method uses existing earthquake data in China. The variables are the lithology and formation structure, slope height, slope angle, and earthquake and rainfall parameters. The method was verified for seismic landslide prediction in western China. The stability of a slope can be expressed as

$$H = (S_{\text{i}} \times S_{\text{a}} \times S_{\text{h}} ) \times (T_{\text{s}} + T_{\text{p}} ),$$
(1)

where \(S_{\text{h}}\) is the influence coefficient of slope height, \(S_{\text{i}}\) is the condition coefficient of the lithology, \(S_{\text{a}}\) is the influence coefficient of the slope angle, \(T_{\text{s}}\) is the strength coefficient of the earthquake, \(T_{\text{p}}\) is the strength coefficient of the rainfall, and H is the stability of the slope. If H < 264, the slope is stable; if H > 396, the slope is unstable. When H is between 264 and 396, the stability of the slope is unknown.

Equation (1) indicates that earthquake-induced landslides are caused by a combination of external factors (earthquakes, rainfall) and internal factors (slope height, lithology, and slope angle). Finally, a comprehensive index value that determines the stability of the slopes in the study area is calculated and the critical value for an earthquake-triggered landslide is determined.

3 Case analysis

The 20-km segment of the Tongmai-105 Daoban was selected as our study area, starting east of the Tongmai bridge and ending in the gang’s former site. This is an important part of the Sichuan–Tibet highway, and it is an existing route reconstruction project. The section of the road is located in the deep-cutting gorge area of Southeastern Tibet. The region experiences a high level of earthquake activity, the neotectonic movement is strong, and the geological structure is complex. The rock is fragmented, loose, and unstable and rainfall is abundant. A maritime glacier spans a wide area, and the river and water systems are continuously evolving, with strongly down-cutting valleys and gullies. The road section is highly prone to geological disasters. It is an ideal study area of earthquake-induced landslides. The image data for this region were provided by the Resources Satellite Three and GF-1 satellite. The image accuracy is 1:25,000.

3.1 Division of slopes threatening the route in the study area

A total of 804 slopes were identified in the study area by using the hydrological analysis method. The average aspect of all the grids was calculated for each slope. The angle between the main axis of the slope and the route was evaluated. Any slope that threatens the routes by the second development of the ArcGIS program was identified. Finally, we generated the map of the slope zones that threaten the road, as shown in Fig. 3. The yellow (1) and red zones (2) in Fig. 4 denote the slopes that have a high risk of forming landslides along the route on the east and west sides of the river, respectively, while the white slopes represent slopes that pose little or no threat to the route.

Fig. 4
figure 4

Zoning map of the slopes threatening the route

Full size image

3.2 Assessment of seismic landslide risk in study area

3.2.1 Climatic characteristics of the study area

The climate of the study area is mild and humid. The annual mean temperature is between 8 and 13 °C and the annual precipitation is between 1,000 and 1,400 mm. The rainy season lasts from April to October and the dry season is from November to March. The precipitation during the rainy season accounts for 85.5 % of the total annual precipitation. The rainy seasons are longer than the dry seasons and most of them are moderate. Heavy rain is infrequent and rainstorms do not exceed a daily precipitation of 50.0 mm.

3.2.2 The average angle and height of each slope

The average angle (Fig. 5) and the relative height of each slope unit were calculated in the second development of ArcGIS5.

Fig. 5
figure 5

The average angle of the slopes

Full size image

3.2.3 Lithology and near-field seismotectonics of the study area

According to geological data provided by department of transportation of Tibet, the bottom of the landslide is covered by loose accumulation masses of Quaternary Holocene (colluvium layer (Q c+dl4 ), a landslide accumulation layer of Quaternary Holocene (Q del4 ), alluvial layer of Quaternary upper Pleistocene (Q al+pl3 ), debris flow accumulation layer of Quaternary Holocene (Q sef4 ), and alluvial layer of Quaternary Holocene (Q 2al+pl4 ). The underlying bedrock consists mainly of the Presinian Gandise rock group (AnZgd) and Presinian Barwa rock group, and the Ani bridge schist group (AnZa), as shown in Fig. 6.

Fig. 6
figure 6

Seismotectonic map of the region surrounding the Tongmai-105 Road section

Full size image

In the study area, seismic waves with Ms ≥ 4.7 were recorded 115 times, seismic waves with 4.7 ≤ Ms ≤ 4.9 26 times, seismic waves with 5.0 ≤ Ms ≤ 5.9 70 times, seismic of Quaternary Holocene waves with 6.0 ≤ Ms ≤ 6.9 16 times, seismic waves with 7.0 ≤ Ms ≤ 7.9 twice, and seismic waves with Ms ≥ 8.0 once. The earliest recorded earthquake is the Langxianzari 4.75 earthquake in 1331 AD. The largest earthquake (Ms 8.6) occurred at Motuo, Tibet in 1950 [19, 20]. The level of seismic activity in this area is high. The earthquakes are very unevenly distributed spatially; most of the major earthquakes are located in the fault zone. There is a possibility of an earthquake of magnitude ~7.0 occurring in the vicinity of the Motou 8.6 earthquake in the next hundred years but the possibility of an earthquake with a magnitude Ms7.5 is small. There are many records of historical earthquakes that occurred in the area surrounding the study site, some of them of small magnitude; this has affected the geology of the region. Modern earthquakes are more frequent, smaller in magnitude, and have a weaker impact on the region. The earthquake with the highest intensity in the study area was the 8.6 Motuo 8.6 earthquake in 1950; the smallest distance to the field region was only 125.4 km. The near-field seismotectonic region is shown in Fig. 6.

3.2.4 Estimation of earthquake-triggered landslides

We used the method of Ding et al. [18] to produce the hazard-zone map of seismic landslides in the study area (Fig. 7). A total of 126 slopes were found to be a threat to the route. Of these, 58 slopes were predicted unstable, 17 slopes were possibly unstable, and 51 slopes were estimated as stable.

Fig. 7
figure 7

Seismic landslide hazard-zone map

Full size image

4 Verification of earthquake-induced landslide hazard zones

The most accurate way to assess landslides and land collapse in the study area is to perform a detailed geological field survey. To verify the accuracy of our seismic landslide hazard zonation map, we compared it with the geological survey reports produced for the renovation reconstruction project in this segment of the route (geological exploration report on this segment provided by the Communications Department of Tibet). The geological survey found 12 landslides posing a great threat to the route along the existing route of the reconstruction project in the Tongmai-105 Road segment. The scale and causes of the landslides are shown in Table 1. There were 7 collapse sites along the project route; their main characteristics are shown in Table 2. In addition, 4 larger landslides were found along the route reconstruction project scheme A. The 23 collapses and landslides mentioned above were marked on the route and compared with our hazard estimation map, as shown in Fig. 7.

Table 1 Landslide along the renovation and reconstruction project in the Tongmai 105 Road segment
Full size table
Table 2 Collapses along the renovation and reconstruction project in the Tongmai-105 Road segment
Full size table

Figure 7 shows that of the 23 collapse and landslide sites identified in the geological exploration report, 19 were identified in our assessment as threats to the route. Only 2 landslide sites (H3, AH1) and 2 collapse sites (B2, B4) were not included in our hazard-zone map as threats to the route. In those sites, the angle between the slope and the route setting is outside the 60–120° range (Sect. 2.2 above). Therefore, they pose a low threat to the route and were excluded from our hazard prediction. Of the remaining 19 collapses and landslide sites, only one landslide (H7) was evaluated incorrectly in our method. The other 18 sites were judged to be unstable in an earthquake, in agreement with the geological survey findings. It should be noted that although not all the collapse and landslide sites identified in the survey were triggered by earthquakes, they are in a dangerous state under static forces. Therefore, these 23 landslides are considered unstable in the event of an earthquake.

Thus, the rapid assessment of seismic landslide hazard zones is a valuable tool for transport route planners and engineers. Its estimation accuracy will be further improved with higher-precision remote sensing image data.

5 Conclusions

Using satellite remote sensing technology, the landslide hazard in rugged mountainous region with active tectonic movement can be well predicted for the determination of new routes and the reconstruction of existing routes. The seismic-triggered landslide hazard estimation method presented here is especially useful for route location in rugged and dangerous mountain region.

  1. 1.

    A great threat to the safe operation of the route exits if the main axis of the landslide and the route aspect intersect at a large angle. According to the field survey of the study area, the most serious damage occurs when the angle between the main axis of the slope and the axis of route is between 60° and 120°. A map of the slopes that pose a threat to the route can be obtained by identifying the high-risk slopes based on the second development of ArcGIS. By using the method of Ding et al. [18], combined with map of the slopes that threaten route, a hazard-zone map of seismic landslides is obtained. The results were verified by comparing with the results of a geological survey report, confirming its high accuracy.

  2. 2.

    To conduct a feasibility study for railway or highway routes, the presented method can quickly generate a seismic landslide hazard-zone map as a guideline for preliminary route selection. A safety corridor can be rapidly selected and workers in the field can clear the selected route. A comprehensive geological survey can then be carried out along the route and more detailed surveys are performed if high-risk sites are identified. Thus, this method can considerably reduce the project costs as well as save time. Moreover, it will contribute to the safety of the operation by preventing landslide and collapse accidents, and losses due to design changes will be significantly reduced.