甘肃科学学报

为了探究青藏高原地区滑坡位移特征并掌握山坡的安全状况,采用AdaBoost算法研究斜坡位移、降雨量以及土壤含水率之间的关系,并进一步对斜坡变形位移进行预测。结果表明:监测区域每年的降雨量呈现倒V形分布,斜坡位移量整体呈现增大趋势;每年二、三月气温升高,受地表积雪积冰融化影响,不同深度的土壤含水率与降雨量的关系复杂。斜坡位移与土壤含水率呈现正相关关系;斜坡不同位置位移受不同深度土壤含水率的影响程度不同。基于监测数据训练AdaBoost回归模型以预测斜坡位移的方法取得了不错的效果,能够对未来15 d内的位移进行预测,研究成果能够为称多县的斜坡位移预测与安全评估提供参考。

In order to explore the characteristics of landslide displacement in Chengduo County of QinghaiTibet Plateau and grasp the safety status of slopes,AdaBoost algorithm was used as a new idea to analyze the relationship between cumulative displacement of slopes,rainfall and soil water content at different depths.The results show that the annual rainfall in the monitoring area presents an inverted V-shape distribution,and the displacement of the slope shows an overall increasing trend.Due to the melting of snow and ice on the surface in February and March every year,the relationship between the soil water content at different depths and the rainfall is unclear.The displacement of slope is affected by soil water content,showing a significant positive correlation.The displacement at different positions of slopes is affected differently by soil water content at different depths.The AdaBoost regression model trained on monitoring data has achieved good results in predicting slope displacement,and can predict displacement within the next 15 days.The research results can provide reference for slope displacement prediction and safety assessment in the Chengduo County area.