利用高分辨率无人机航拍影像，结合基本地质资料，分析了影响2014年 8月3日鲁甸M_S6.5地震震后崩塌滑坡分布的主要因素，使用M5'模型树算法建立了崩塌滑坡密度与其影响因子间的分段线性模型，并检验了该模型的预测性能。结果表明，地震诱发的崩塌滑坡分布受断层距、岩土体结构强度、坡度、植被条件等的影响，其中，断层距、岩土体结构强度及坡度等为主要影响因素；崩塌滑坡易发生在结构破裂区及坡度为38°~50°的区域，其分布密度随断层距的增加而减小；利用M5'模型树算法建立的模型体现出崩塌滑坡分布与其影响因子间复杂的非线性关系，模型检验结果显示，理论模型与实际关联函数间的相关系数达到0.88，因此，可利用该模型预测地震诱发的崩塌滑坡的分布。

An earthquake with M_S6.5 occurred on August 3,2014,in Ludian,Yunnan Province,China. The Ludian earthquake resulted in enormous casualties,and also caused a large number of landslides. We use the high-resolution aerial images with basic geological data of aerial area in the analysis of the distribution of landslides induced by the earthquake with the main affecting factors of structural strength of rocks,slopes,vegetation conditions and distance to the fault. A piecewise linear model was established by the M5' model Tree method which has four explicit piecewise linear indexes including the density of landslides,distance from the fault,structural strength of rocks,slopes and vegetation conditions. The results show that the landslides are mainly affected by distance to the fault,slopes and structural strength of rocks. The density of landslides is attenuated with the increase of distance to the fault. Earthquake-induced landslides form easily in the structural fracture zone and slope with 38°-50°. The established model approaches the nonlinear relations earthquake-induced landslide distribution with those factors. The correlation coefficient between the model and the actual distribution of landslides density is 0.88. It denotes that landslides density is determined by the factor referring to the index. Therefore,model might be used to predict the distribution of earthquake-induced landslides.

P315

中国地震局地球物理研究所中央级公益性科研院所基本科研业务专项（DQJB14B10、DQJB15C04）和国家自然科学基金项目（41404048）联合资助