关键词: 玛多地震，地震滑坡，危险性分析，Newmark模型，判别分析法

Abstract: The earthquake-triggered landslides and slope failures are common phenomena during strong earthquakes in mountain areas, which causes more casualties and economic property losses than the disaster itself in many cases. In order to reduce the loss of earthquake landslide disaster, scientists have formed a variety of prediction and evaluation methods of earthquake landslide hazard based on different theories and models through long-term research. The MS7.4 earthquake, which occurred at 2:04 a.m. on 22 May 2021 in Maduo,Qinghai (34.59 °N, 98.34 °E), provided an opportunity to test the validity of the different models. On the one hand, based on the simplified Newmark displacement model, the sensitivity of seismic landslide in Maduo earthquake area is calculated . Furthermore, the seismic landslide risk is evaluated by combining with the seismic intensity distribution map after Maduo earthquake. On the other hand, based on the discrimination analysis method, the empirical model obtained from the Niigata earthquake in Japan is used to predict the earthquake landslide in Maduo earthquake area. The research results show that: based on the rapid assessment of earthquake induced landslide risk by simplified Newmark displacement model, the potential high-risk areas are mainly concentrated in the intensity area of Ⅷ ⅨⅩ which is greatly affected by the intensity of ground motion. On the whole, with the weakening of the impact of ground motion, the landslide risk decreases gradually, which is in good agreement with the objective reality. As an empirical model, discrimination analysis method is relatively dependent on a specific environment. When it is used out of its own environment, it is necessary to verify the universality of empirical formula, re-understand the relationship between various impact factors, and adjust the weight of each factor. The difference between the two methods in the prediction results is mainly distributed in the Ⅵ degree area. Above the study area, the risk zoning obtained by the two methods is generally consistent. Due to the differences in the research models adopted by the two methods, there are some differences in the distribution of seismic landslide hazard areas with different risk levels in the prediction results, especially in the Ⅵ intensity region, which is quite different: Ⅵ Intensity area is wide, mountainous area is large, and most of the areas have steep slope. As a result, the discriminant analysis results in this area are more influenced by slope and curvature value, so that there are more highly dangerous areas in the prediction results. However, the simplified Newmark method is greatly affected by the ground motion. Because this region is far away from the epicenter, the impact of ground motion is weak, so the main prediction results in this region are low risk areas. However, in the study area of Ⅶdegree and above, the risk zoning of the two methods was generally consistent, and the prediction effect was good. In seismic landslide risk assessment based on discriminant analysis method, objective and complete landslide samples need to be fully analyzed, which is also a problem faced by the prediction method based on empirical model. As a physical model, Newmark model does not depend on the specific environment, although it has the problem of the accuracy of input parameters, so it is more objective and reasonable in the calculation results.

Key words: the Maduo earthquake, earthquake-triggered landslides, Hazard analysis, Newmark model, discrimination analysis method.