Spatio-temporal Characteristics And Susceptibility Of Post-seismic Landslides Along The High Seismic Intensity Area In The Eastern Tibetan Plateau

Landslide is one of the major geologic hazards induced by strong earthquakes in the mountainous region with steep landforms.The strong earthquake can not only trigger many coseismic landslides,but also cause surface deformation and cracks,geomorphic change,and form a large number of loose materials,making the post-seismic landslide activities to last for a long time.During the post-seismic reconstruction,the active landslides have become significant disasters threatening human life and property.It is essential to analyze the common characteristics of post-seismic landslide activity,evaluate their susceptibility,and explore the change law of susceptibility for disaster prevention and reduction.However,the post-seismic landslides activities are complex.Previous studies focused on the coseismic landslides,and the studies on post-seismic landslides is less.They could not establish an efficient evaluation system of time-variant post-seismic landslides susceptibility.Lack of effective research explores the common spatio-temporal distribution characteristics and general laws on post-seismic landslide susceptibility.Therefore,this study focuses on the earthquake-related landslides along the high seismic intensity area in the eastern Tibetan Plateau.It tries to establish the accurate evaluation of post-seismic landslide susceptibility and summarize the common spatio-temporal distribution characteristics and general laws on post-seismic landslide susceptibility.It is one of the most complex and intense seismicity regions,with several large historical seismic events in the eastern Tibetan Plateau.Strong earthquakes and steep terrain induced many landslides in this area.Due to the dense vegetation,the damage caused by landslides is significant and easy to identify in optical remote sensing images.It is helpful to establish the post-seismic landslide dataset and provide a natural laboratory for studying the characteristics of post-seismic landslide distribution and the laws of post-seismic landslide susceptibility.In this study,the automatic classification,visual interpretation,and field investigation combined with integrated “Space-Ground” remote sensing monitoring technology is implemented to extract landslides in each image and establish a rapid and accurate identification system of landslides.Then,the study used change detection of the extracted landslide range in different phase images to acquire the actual post-seismic landslides in order to build the multi-temporal post-seismic landslide dataset.The buffer analysis tool generated a multi-level buffer zone from the edge of the previous landslide.Spatio-temporal distribution characteristics of post-seismic landslides are explored by counting the landslides in different buffer zones.According to the characteristics,this study obtains the distance as a new influence factor to quantify the legacy effect.The landslides with different geological conditions,which are associated with the Jiuzhaigou earthquake in Sichuan and Minxian earthquake in Gansu,are chosen as the test objects to conduct comparative experiments.At first,I+LR model was selected as the primary model for establishing an improved model.Then,the primary influencing factors such as seismic,topographic,geological,and human activity data,combined with the distance factor,were input into the primary model,establishing the improved post-seismic time-variant landslide susceptibility model.The landslides associated with the Jiuzhaigou earthquake and the Wenchuan earthquake in the eastern Tibetan Plateau are taken as typical case studies.The improved landslide susceptibility model was used to calculate their post-seismic susceptibility.And the transfer matrices of post-seismic landslide susceptibility have been implemented to analyze the transfer trend and proportion of post-seismic landslide susceptibility.The common characteristics of post-seismic landslides distribution and change laws on post-seismic landslide susceptibility are summarized and discussed in detail.Main results and conclusions are described as follows:(1)For selecting influence factors and primary model,seismic,topography,geology,and human activities are reasonable for conventional susceptibility models and can obtain high accuracy.They can be used as the basic influencing factors.Accuracy of the coupling model in the conventional model is generally higher than that of the simple model,and the landslide susceptibility distribution map is more reasonable for the same landslide dataset.This study indicated that I+LR model can acquire the best comprehensive effect in the common coupling models.(2)For spatio-temporal distribution of post-seismic landslides,the post-seismic landslides are mainly caused by expansion of the original landslides,which are mostly distributed within a distance of 200 m from the original landslides.The post-seismic landslides become gradually close to the original landslides with time.New post-seismic landslides have a significant spatial correlation with the original landslides,which indicates that the distance is a critical factor in measuring the legacy effect of original landslides for the future landslides.(3)For improvement of the susceptibility model,distance is a critical factor of measuring the legacy effect,and it can be obtained by simple statistical analysis.This study built an improved model by inputting the distance and the basic influencing factors into the primary landslide susceptibility model.It can reasonably calculate the post-seismic landslide susceptibility and enhance the accuracy.Compared with the path-dependent method,it is a simple approach and is suitable for the most post-seismic multi-temporal landslide datasets,including datasets with simple attributes.(4)For the characteristics and laws of post-seismic landslide susceptibility,the correlation between the post-seismic landslide susceptibility and seismic factors,such as fault and PGA,gradually became weak with time.It is progressively affected by unstable slope deposits and rainfall.The high and very high landslide susceptibility areas shrink rapidly towards the landslide concentration areas.The post-seismic landslide susceptibility transferred to lower susceptibility in early-stage after the earthquake,with a high transfer proportion.Transfer speed is slowing down with time,but the transfer trend is unchangeable.If it is disturbed by external factors,the transfer of post-seismic landslide susceptibility will fluctuate in a short period.For a long-term scale,it still trends to decrease.In general,an efficient post-seismic landslide identification method has been established to enhance the extraction accuracy in this study.The improved time-variant susceptibility model built by this study is characterized by the high accuracy,wide application,and easy implementation.Furthermore,the improved model was applied to calculate the post-seismic landslides susceptibility of two strong earthquakes along the high seismic intensity area of the eastern Tibetan Plateau.It demonstrated that this new improved model is helpful to conclude the common spatio-temporal distribution characteristics and the general change laws of post-seismic landslides susceptibility.Above-mentioned results provide important information for the treatment of post-seismic landslides and eco-environmental conservation of the large earthquake-hit regions in the eastern Tibet.Additionally,those large-scale post-seismic landslides need a long-term cycle for landslide restoration.The slopes covered by vegetation in ecological restoration are still easy to slide affecting by abnormal factor like rainfall,which need to be monitored,evaluated and treated for a long time.According to the results of spatio-temporal change for post-seismic landslides,suggesting that the maximum protection distance can be set as ca.200 m,and the monitoring duration may last for more than 10 years.