清江神女洞岩溶洞穴沉积物的磁学研究

通过对清江神女洞岩溶沉积物的磁学研究,得出有关该洞沉积物形成环境的下列判断：岩溶洞穴沉积物形成于早更新世至中更新世;岩溶洞穴沉积物形成于湿暖、干冷交替的环境;岩溶洞穴沉积物形成于ＮＥ５６°流向的古水流中。     

The Loess in the Qinling Mountains and Its Records of Palaeoenvironmental Changes

The Fengzhou loess section is very typical in the Qinling Mountains. The section, which is about 82 m thick and underlain by Neogene red clays, consists of 33 layers of loess and 33 layers of palaeosol. It covers the Brunhes normal polarity zone and Matuyama reversed polarity zone, and the B/ M boundary is located in the middle part of loess layer 8 (L8). The loess of the Matuyama reversed polarity zone records the Jaramillo, Olduvai and Reunion normal polarity subchrons. The boundary between the Matuyama reversed polarity zone and Gauss normal polarity zone (M / Ga) appears on the lithological boundary between loess and Neogene red clays. Loess accumulation in the Fengzhou section started before 2.48 Ma B.P. Based on the stratigraphical structure, the material composition and magnetic susceptibility curve of the Fengzhou loess section, the palaeoclimatic changes during the last 2.48 Ma in the Qinling Mountains are subdivided into 66 cold-dry and warm-humid stages, equivalent to 33 climatic cycles. The     

层次分析-熵值定权法应用于山区城镇地质灾害易发性评价

我国西南山区很多城镇面临着崩塌、滑坡等地质灾害的威胁,开展山区城镇地质灾害易发性评价研究具有重要意义。本文对重庆市武隆区羊角场镇进行了地质灾害易发性评价研究,在确定区域易发性评价因子的基础上,综合层次分析法和熵权法两种方法确定评价因子的权重,最后采用易发性指数并基于ArcGIS对研究区进行易发性评价和区划。易发性分区结果与已有的地质灾害分布有较好的对应关系。采用的组合赋权法先用层次分析法确定主观权重,然后又结合熵值法对权重进行修正,在一定程度上减少了主观影响,对地质灾害易发性的量化更加准确与合理。本文的研究思路可为山区城镇地质灾害的风险评估和防治提供参考。     

基于深度学习的滑坡灾害易发性分析

滑坡灾害成因机理复杂、影响因素众多,深度学习作为当前人工智能领域的热点,能够更好地模拟滑坡灾害的形成并准确预测潜在的斜坡。为了挖掘深度学习在滑坡易发性的应用潜能,本文构建了一维、二维和三维的滑坡数据表达形式,并提出3种基于卷积神经网络模型（Convolutional Neural Networks, CNN）的滑坡易发性分析处理框架：基于CNN分类器、基于CNN与逻辑回归的融合和基于CNN集成,最后以江西省铅山县为研究对象进行验证,结果表明：所有基于CNN的易发性模型都能够获得准确且可靠的滑坡易发性分析结果。其中,基于二维数据的CNN模型在所有单分类器中预测精度最高,为78.95%。此外,二维CNN特征提取能够显著提升逻辑回归的预测精度,其准确率提升7.9%。最后,异质集成策略能够大幅度提升基于CNN分类器的滑坡预测精度,其准确率提升4.35%~8.78%。     

内蒙古中部地区岩矿石磁化率特征

为了解内蒙古中部的锡林浩特、巴林左旗等地区岩矿石磁性特征,更好地对该地区的航磁异常进行综合解释,使用航遥中心研制的ZH-1磁化率仪,于2009~2011年间先后四次在该地区进行了野外岩矿石磁化率实测。并在统计、分析实测磁化率数据的基础上,研究了该地区岩矿石磁化率分布情况,分别叙述了区内沉积岩、侵入岩、火山岩、变质岩、矿石及围岩等各种类型岩矿石的磁化率特征,对岩矿石磁化率与航磁、矿产间的联系进行探讨,为磁法勘查解释、地质找矿提供了重要依据。     

人工神经网络和决策树模型在滑坡易发性分析中的性能对比

机器学习模型广泛应用于区域性滑坡易发性分析。模型的选择关系到评价结果的可信度、准确率和稳定性。现有滑坡易发性分析模型对比研究侧重模型的预测精度。模型的稳定性和数据量敏感性对机器学习模型的性能评估同样非常重要。本文以福建省南平市蔡源流域为研究区,以四川省绵阳市北川县为验证区,从预测精度、稳定性和数据量敏感性3个方面深入对比BP（Back Propagation）人工神经网络模型和CART（Classification and Regression Tree）决策树模型在滑坡易发性分析中的效果,主要结论如下:① 在逐渐增加一定数量训练样本的过程中,BP人工神经网络模型预测精度的增长率更高。在蔡源流域内,当训练样本数量增加10 000时,BP人工神经网络模型的预测精度上升5.22%,CART决策树模型的预测精度上升2.11%。② BP人工神经网络的预测精度高于CART决策树模型,且较为稳定。在100组数据集上,BP人工神经网络模型验证集预测精度的均值和验证集滑坡样本预测精度的均值分别为81.60%和84.86%,高于CART决策树模型的72.97%和76.59%。与此同时,BP人工神经网络模型对应预测精度的标准差分别是0.32%和0.37%,小于CART决策树模型的0.35%和0.67%。③ BP人工神经网络模型分析的滑坡易发区相比CART决策树模型,更接近实际滑坡的空间分布。最后,北川县的验证实验也出现了相同的现象。     

Climatic implications of loess deposits from the Beijing region

Variations in magnetostratigraphy, pedostratigraphy, grain size and magnetic susceptibility of the loess deposits near Beijing have been studied at two sections. The sections are about 400 km east of the main loess deposits in China, have a maximum thickness of 100 m and extend back to 1.1 Ma. The sequence consists of 14 loess–palaeosol couplets (S0‐S14), which correlate well with sequences in the Loess Plateau. Susceptibility records from the sites near Beijing are comparable to the Xifeng, Luochuan and Baoji sections located in the middle part of the Loess Plateau; however, the down‐core variations in the grain size in the Upper Lishi Formation exhibit some differences. The median grain size increases by about 25–30 µm from L4 to L2, with the sandy grains (>63 µm) increasing from 10–20 wt% to 40–50 wt% . This implies that the depositional environment of the dust sources in the Beijing loess section is different in some aspects from the Loess Plateau. The Beijing loess may have had a different dust source than the Loess Plateau. Copyright © 2001 John Wiley & Sons, Ltd.     

国道213汶川—松潘段滑坡隐患遥感识别与易发性评价

国道213汶川—松潘段位于强震山区,滑坡灾害频发,每年都会因滑坡灾害导致交通中断,迫切需要查明沿线滑坡隐患的空间分布,并对其易发性进行评价。利用光学遥感和InSAR综合遥感技术对沿线滑坡隐患进行识别,共识别滑坡隐患288处,其中InSAR探测出有形变的滑坡隐患点27处。以识别出的滑坡隐患为评价样本,选取高程、坡度、坡向、地表曲率、工程地质岩组、归一化植被指数(normalizied difference vegetation index,NDVI)、距道路距离、距河流距离、距断层距离等9个影响因素作为评价因子,采用Logistic回归模型评价该沿线滑坡隐患易发性,评价结果可为国道213汶川—松潘段滑坡灾害防治提供参考。     

The anisotropy of loess magnetic susceptibility in the northeastern fringe of Qinghai-Xizang Plateau as an indicator of palaeowind direction

Estimates of the palaeo-subaerial wind direction were studied systematically for the first time by using the anisotropy of loess magnetic susceptibility (AMS) measurements in the northwestern China. One hundred and forty undisturbed oriented aeolian loess samples were collected from Lanzhou, Linxia and Wudu areas for AMS measurements, which indicated the subaerial wind directions were not the same while the loess deposited. From the Early Pleistocene to Middle Pleistocene till Late Pleistocene, the wind direction experienced an anticlockwise rotation in the studied area. We suggested this change was related to the uplift of the Qinghai-Xizang Plateau and the adjustment of current and landform effects.     

Contribution of domain rotation to the magnetic susceptibility of spherical two-domain particles of magnetite

Summary. Starting with a simplified picture of the presumed magnetization configuration in a two-domain sphere of magnetite, the initial susceptibility due to rotation of the magnetization in the domains has been calculated. In particles between 100 and 300 nm in diameter the calculated susceptibility for randomly oriented particles drops rapidly from about 6 to a little over 3, which is the value to be expected for large multidomain particles. The results agree well with the available experimental results and imply that in magnetite particles in this size range the low field susceptibility is dominated by the domain rotation and that wall motion is effectively inhibited.