Study On Semi-automatic Extraction Of Landslide And Disaster Process In The Past 34 Years In Eastern Himalayan Syntaxis

As one of the most active areas in the world,Eastern Himalayan Syntaxis has frequent geological disasters such as landslides.The disaster chain of landslide-dam-flood is not only an important process of shaping the surface morphology,but also a serious threat to people’s lives and property safety.Landslide is the source of this disaster chain.Clarifying its spatial distribution characteristics,influencing factors and susceptibility assessment are the basis for assessing the degree and scope of disaster chain process,and also an important scientific basis for preventing disasters and reducing damages.Limited by previous remote sensing technology and computer technology,it is difficult to quickly establish a reliable landslide inventory in a large space,which restricts the complete and continuous prediction and evaluation of this disaster chain process.With the enrichment of remote sensing images and the development of machine learning technology,it is possible to construct a reliable landslide list,predict the disaster process and conduct the hazard assessment.In this study,a change detection landslide extraction method based on difference image improved by particle swarm optimization algorithm is used to construct a list of historical landslides from 1987 to 2021 in the eastern tectonic knot,and the frequency,amplitude and spatial distribution characteristics of landslides are obtained.The control factors of landslide are found and the weight assignment is carried out according to their influence degree.The landslide susceptibility model is constructed based on the analytic hierarchy process,and the susceptibility evaluation is carried out to obtain the landslide susceptibility map of the study area.According to the frequency,amplitude and spatial distribution characteristics of landslide,the risk assessment of landslide dam is constructed based on the morphological blocking index.Based on the size of the dammed lake,the height of the dam and the gradient of the downstream section of the dammed lake,the risk level of the landslide dammed lake is evaluated.The map of the damming risk level and the map of the landslide dammed lake risk level in the study area are obtained.The scientific suggestions for preventing disasters and reducing damages in the study area are proposed.The main conclusions are as follows:(1)An inventory of historical landslides in Eastern Himalayan Syntaxis is constructed to obtain the spatial distribution characteristics and erosion rate of landslides.1389 landslides are extracted,with an area of 46.8 km~2,accounting for0.09%of the total area of the study area,and a total volume of about 1.0×10~9m~3.The overall distribution of landslides in the study area is characterized by more in the east and south and less in the west and north.The average landslide erosion rate in the study area is 0.47 mm/a,and the maximum erosion rate is 43.88 mm/a,which appears in the Yarlung Zangbo Grand Canyon.There is a spatial consistency between the erosion rate of the landslide and the river incision rate,indicating that the landslide is an important process for the mountain to respond to the rapid tectonic uplift in Eastern Himalayan Syntaxis.(2)Based on the analytic hierarchy process,the landslide susceptibility model of Eastern Himalayan Syntaxis is constructed,and the landslide susceptibility map of the study area is drawn.The results show that the landslides in the study area are more likely to occur in the elevation range of 1500-2000 m,the slope range of 35-40°,the slope aspect of southeast slope,the lithology of metamorphic rock,the distance of3000 m from the fault,the peak ground acceleration of 0.3 g,the average annual temperature of more than 5℃ and the average annual precipitation of more than 800mm.The landslide susceptibility model constructed in this study can judge the landslide susceptibility in the study area well,and the area under the ROC curve of the prediction results reaches 0.761.(3)The risk of landslide damming and the risk grade of dammed lake are evaluated.The results show that 7.8%of the landslides in the study area may lead to damming when reactivated,including 13 very high dam risk and 3 high dam risk landslides.Among the very high and high damming risk landslides,11 dams may be formed in the Yarlung Zangbo River mainstream,and 2 of the dammed lakes are of high risk and 2 are of medium risk.Two dams may be formed in the Yigong Zangbu River,and one of the dammed lakes is of extremely high risk,one is of medium risk.Two dams may be formed in Palong Zangbu River,and one of the dammed lakes is of a moderate risk;one dam may be formed in other small tributaries.There are 42landslides with damming risk in the 25 km long river channel of the Yarlung Zangbo River,with a density of 1.7 landslides/km.(4)According to the characteristics of landslide disasters in the study area,scientific suggestions for disaster prevention and mitigation are proposed.(1)Strengthen landslide monitoring in wide valley-canyon conversion Area.(2)Strengthen the monitoring of the stability of large landslide deposits.(3)Carry out comprehensive assessment of disaster-causing factors before construction,and reduce the influence of landslide disaster by river dredging and slope stabilization.In summary,this paper develops a standard process for automatic extraction of landslides based on particle swarm optimization algorithm,and establishes a reliable list of landslides in the past 34 years in Eastern Himalayan Syntaxis.The frequency,amplitude,spatial distribution characteristics and prone areas of landslides are obtained.Based on the frequency and amplitude of landslide,the risk assessment of landslide damming and the risk level of dammed lake are developed.It is found that the Yarlung Zangbo River Grand Canyon is the area with the highest risk level of landslide damming and dammed lake,and key monitoring should be envolved.It is an effective means of preventing disasters and reducing damages to promote the improvement of landslide monitoring and early warning system,improve the society’s emphasis on landslide and its disaster process,carry out science education for the entire people,and improve the system by residents’self-control of disaster.