吉林建筑大学学报

新疆额敏县布尔阔台河下游沿岸发育多条沟谷型泥石流，但受限于区域气候较干旱与人类活动较少，泥石流灾害研究基础薄弱，相关地质数据匮乏，物源识别和危险性定量分析存在较大困难。传统二维遥感方法难以满足复杂地形下的物源识别精度需求，成为制约区域灾害评估精度的技术瓶颈。通过野外调查、卫星遥感解译与无人机倾斜摄影等手段，获取了研究区16条泥石流沟(Z01～Z09、Y01～Y07)的精细地质信息，结合无人机构建的高精度三维实景模型，开展了物源类型划分与储量估算；进一步选取15个泥石流特征因子，判定各沟易发性等级，并引入灰关联法，结合10个危险性评价指标(含2个主要因子与8个次要因子)，量化评估各泥石流沟的危险度。结果表明，区域内沟谷普遍具备孕灾地貌与物源条件，其中14条沟为中等易发，2条沟为轻度易发；但总体危险度均属于轻度等级。本研究创新性地提出了“无人机三维建模+灰关联评价”融合技术路径，为半干旱区沟谷型泥石流的快速识别与科学评估提供了新方法。

A number of valley-type debris flow gullies have developed along the lower reaches of the Buerkuotai River in Emin County, Xinjiang. However, due to the region's semiarid climate and limited human activity, data scarcity and weak debris flow development pose challenges to risk evaluation. In particular, conventional 2D remote sensing methods are insufficient for accurately identifying source materials in complex terrains. Through field investigation, satellite image interpretation, and unmanned aerial vehicle(UAV) oblique photography the fine geological information of 16 debris flow gullies(Z01～Z09 and Y01～Y07)in the study area was obtained. Combined with the high-precision 3D real-lifemodel constructed by UAV, the classification of provenance types and reservesestimation were carried out, and 15 debris flow indicators were selected to assess susceptibility levels, and ten risk assessment factors(including two primary and eight secondary factors) were used in a grey correlation-based evaluation model. Results indicate that 14 gullies are moderately susceptible and two are mildly susceptible, with all classified as mildly dangerous overall. This study proposes an innovative methodology that combines UAV-based 3D modeling and grey correlation analysis, offering a novel approach for rapid debris flow assessment in semiarid mountainous regions.