It is of crucial importance to investigate the spatial structures of ancient landslides in the eastern Tibetan Plateau’s alpine canyons as they could provide valuable insights into the evolutionary history of the lan...It is of crucial importance to investigate the spatial structures of ancient landslides in the eastern Tibetan Plateau’s alpine canyons as they could provide valuable insights into the evolutionary history of the landslides and indicate the potential for future reactivation.This study examines the Deda ancient landslide,situated in the Chalong-ranbu fault zone,where creep deformation suggests a complex underground structure.By integrating remote sensing,field surveys,Audio-frequency Magnetotellurics(AMT),and Microtremor Survey Method(MSM)techniques,along with engineering geological drilling for validation,to uncover the landslide’s spatial feature s.The research indicates that a fault is developed in the upper part of the Deda ancient landslide,and the gully divides it into Deda landslide accumulation zoneⅠand Deda landslide accumulation zoneⅡin space.The distinctive geological characteristics detectable by MSM in the shallow subsurface and by AMT in deeper layers.The findings include the identification of two sliding zones in the Deda I landslide,the shallow sliding zone(DD-I-S1)depth is approximately 20 m,and the deep sliding zone(DD-I-S2)depth is 36.2-49.9 m.The sliding zone(DD-Ⅱ-S1)depth of the DedaⅡlandslide is 37.6-43.1 m.A novel MSM-based method for sliding zone identification is proposed,achieving less than 5%discrepancy in depth determination when compared with drilling data.These results provide a valuable reference for the spatial structural analysis of large-deepseated landslides in geologically complex regions like the eastern Tibetan Plateau.展开更多
微动HVSR(horizontal-to-vertical spectral ratio)法是一种高效、非侵入式的地球物理探测手段,广泛用于城市地质调查与工程勘探。然而,行人、车辆等瞬态干扰会导致HVSR曲线畸变。现有瞬态干扰剔除方法存在局限:STA/LTA(short-term-aver...微动HVSR(horizontal-to-vertical spectral ratio)法是一种高效、非侵入式的地球物理探测手段,广泛用于城市地质调查与工程勘探。然而,行人、车辆等瞬态干扰会导致HVSR曲线畸变。现有瞬态干扰剔除方法存在局限:STA/LTA(short-term-average over long-term-average)法易误判且调参复杂,手动剔除法效率低,频域窗剔除法仅关注峰值频率信息。为此,本文提出一种基于机器学习的微动HVSR数据干扰信号压制方法。首先通过提取曲线形态特征训练曲线剔除模型,用于剔除HVSR数据中显著偏离平均趋势的离群HVSR曲线;随后提取峰值特征训练峰值识别模型,用于识别曲线中的有效共振峰值;最后结合DBSCAN(density-based spatial clustering of applications with noise)聚类算法对识别出的有效峰值进行聚类与二次剔除,去除频率和振幅异常的峰值所在曲线。曲线剔除模型和峰值识别模在测试集上的F_(1)分数分别为0.967和0.985,均表现出优异的分类性能。实际算例结果表明,本文方法在异常曲线剔除方面相较于STA/LTA法和频域窗剔除法具有更高的稳定性与准确性,处理后的曲线频谱分布更加集中,标准差曲线更加收敛,峰值更加清晰稳定。本文方法在实现高效自动处理的同时,剔除效果与手动剔除法高度一致。展开更多
基金supported by the National Natural Science Foundation of China(42372339)the China Geological Survey Project(DD20221816,DD20190319)。
文摘It is of crucial importance to investigate the spatial structures of ancient landslides in the eastern Tibetan Plateau’s alpine canyons as they could provide valuable insights into the evolutionary history of the landslides and indicate the potential for future reactivation.This study examines the Deda ancient landslide,situated in the Chalong-ranbu fault zone,where creep deformation suggests a complex underground structure.By integrating remote sensing,field surveys,Audio-frequency Magnetotellurics(AMT),and Microtremor Survey Method(MSM)techniques,along with engineering geological drilling for validation,to uncover the landslide’s spatial feature s.The research indicates that a fault is developed in the upper part of the Deda ancient landslide,and the gully divides it into Deda landslide accumulation zoneⅠand Deda landslide accumulation zoneⅡin space.The distinctive geological characteristics detectable by MSM in the shallow subsurface and by AMT in deeper layers.The findings include the identification of two sliding zones in the Deda I landslide,the shallow sliding zone(DD-I-S1)depth is approximately 20 m,and the deep sliding zone(DD-I-S2)depth is 36.2-49.9 m.The sliding zone(DD-Ⅱ-S1)depth of the DedaⅡlandslide is 37.6-43.1 m.A novel MSM-based method for sliding zone identification is proposed,achieving less than 5%discrepancy in depth determination when compared with drilling data.These results provide a valuable reference for the spatial structural analysis of large-deepseated landslides in geologically complex regions like the eastern Tibetan Plateau.
文摘微动HVSR(horizontal-to-vertical spectral ratio)法是一种高效、非侵入式的地球物理探测手段,广泛用于城市地质调查与工程勘探。然而,行人、车辆等瞬态干扰会导致HVSR曲线畸变。现有瞬态干扰剔除方法存在局限:STA/LTA(short-term-average over long-term-average)法易误判且调参复杂,手动剔除法效率低,频域窗剔除法仅关注峰值频率信息。为此,本文提出一种基于机器学习的微动HVSR数据干扰信号压制方法。首先通过提取曲线形态特征训练曲线剔除模型,用于剔除HVSR数据中显著偏离平均趋势的离群HVSR曲线;随后提取峰值特征训练峰值识别模型,用于识别曲线中的有效共振峰值;最后结合DBSCAN(density-based spatial clustering of applications with noise)聚类算法对识别出的有效峰值进行聚类与二次剔除,去除频率和振幅异常的峰值所在曲线。曲线剔除模型和峰值识别模在测试集上的F_(1)分数分别为0.967和0.985,均表现出优异的分类性能。实际算例结果表明,本文方法在异常曲线剔除方面相较于STA/LTA法和频域窗剔除法具有更高的稳定性与准确性,处理后的曲线频谱分布更加集中,标准差曲线更加收敛,峰值更加清晰稳定。本文方法在实现高效自动处理的同时,剔除效果与手动剔除法高度一致。
