基于CatBoost-SHAP模型的滑坡易发性建模及可解释性

文章致力于深入探索滑坡易发性建模中集成学习模型的不确定性和可解释性。以浙江省东部沿海山区为研究对象,利用谷歌历史影像与Sentinel-2A影像,记录了2016年超级台风“鲇鱼”触发的552起浅层滑坡事件。研究首先对连续型因子进行了不分级、等间距法和自然断点法的工况设计,进一步划分为4,6,8,12,16,20级。随后,引入了类别增强提升树模型（CatBoost）以评估不同工况下的滑坡易发性值,再结合受试者曲线与沙普利加性解释法分析,对建模过程中的不确定性和可解释性进行了深入研究,目的在于确定最优建模策略。结果表明：（1） 在CatBoost模型计算中,河流距离成为最关键的影响因子,其次是与地质条件、人类活动相关的因子;（2） 不分级工况下,模型能够获得最高的AUC值,达到0.866;（3）相较于等间距法,自然断点法的划分策略展现出更佳的泛化能力,且模型预测性能随着分级数量的增加而增加;（4）沙普利加性解释法模型揭示了主要影响因子道路距离、河流距离、DEM和坡向对台风诱发滑坡的控制机制。研究成果能够加深对滑坡易发性的理解,提高滑坡预测的准确性和可靠性,为相关地区的防灾减灾工作提供科学依据。     

中-晚始新世渭河盆地古气候变化的红河组磁化率记录

始新世暖期可为理解现代增温过程和机制提供历史相似型。广泛分布于渭河盆地的红河组为中始新世到晚始新世的滨浅湖相-冲积扇堆积,剖面出露厚度超过330 m,盆地沉积中心钻探厚度大于1000 m,为一套红色的碎屑沉积间化学沉积序列。盆地边缘的红河组沉积物含碳酸盐岩、泥岩、砂岩和少量砾岩,沉积中心以泥质灰岩、泥岩和粉砂岩为主。对蓝田支家沟红河组（时代为46~33 Ma）2910个样品的磁化率测试表明,红河组沉积物磁化率值在0~20×10^-8 m³/kg之间,整体偏低,频率磁化率值也相对较低（0~11%）,与中国西北部新生代红层磁化率总体较低相似。高分辨率磁化率测试证实了红河组堆积时期干湿变化的多旋回和阶段性,受到太阳辐射变化影响,并受到下垫面的反馈驱动。红河组泥岩和粉砂岩的磁化率值偏高,而砂岩的磁化率值较低,载磁矿物以赤铁矿为主。盆地周围岩体弱磁性物质的快速输入,降低了红河组粗颗粒沉积物的磁化率值。在中、晚始新世的轨道-构造时间尺度,红河组沉积序列粒度变化受到湖面波动控制。湖面上升时湖滨地区沉积物颗粒偏细,磁化率值高;反之,磁化率值低。因此,磁化率高低变化指示了湖面高低波动,进而指示了气候的干湿变化。分析还表明,中、晚始新世渭河盆地湖面多旋回变化和阶段性变干,与全球温度变化和降温过程可良好对比,指示全球温度降低可能抑制水汽循环,进而驱动始新世东亚内陆干旱的发展。

     

张北汉诺坝玄武岩的岩石磁学性质研究

通过新河口汉诺坝玄武岩的综合岩石磁学测定和分析,确定了该剖面21个熔岩流的磁性矿物是以钛磁铁矿为主,部分含钛较高. 基于磁化率各向异性测定,推断张北汉诺坝玄武岩是由位于北东和北北西两个方向的古火山口喷发而成,这与野外观察一致. 这一结果对探讨火山喷发历史和地球深部过程是有意义的.     

陕西宝鸡地区千阳县地质灾害易发性评价

在对千阳县地质灾害详细调查的基础上,采用定性分析与定量评价相结合的方法,对千阳县地质灾害易发程度进行了评价分区。首先初步定性分析了地质灾害的分布特征和影响因素;其次利用地理信息系统技术,选取斜坡几何形态、工程地质岩组、黄土覆盖层厚度、地质构造和水文地质条件5个方面共28个指标作为评价因子,利用信息量模型对千阳县的地质灾害进行易发性评价分区。评价分区结果表明：高易发区占全区面积的4.89%,主要分布于千河北岸黄土塬区一级支流沟谷两侧的塬边斜坡地区;中易发区占全区面积的24.18%,主要位于千河北岸台塬区一级支流沟谷的两侧,北部山区二、三级支流冲蚀的基岩（白垩系—侏罗系砂泥岩）与黄土的接触区,南部山区局部黄土覆盖的高陡区域;低易发区占全区总面积的41.29%,主要位于千阳北部山区的二、三级河流河道冲蚀区,千河北岸黄土塬区和千河南部的基岩山区。     

Neural networks and landslide susceptibility: a case study of the urban area of Potenza

For those working in the field of landslide prevention, the estimation of hazard levels and the consequent production of thematic maps are principal objectives. They are achieved through careful analytical studies of the characteristics of landslide prone areas, thus, providing useful information regarding possible future phenomena. Such maps represent a fundamental step in the drawing up of adequate measures of landslide hazard mitigation. However, for a complete estimation of landslide hazard, meant as the degree of probability that a landslide occurs in a given area, within a given space of time, detailed and uniformly distributed data regarding their incidence and causes are required. This information, while obtainable through laborious historical research, is usually partial, incomplete and uneven, and hence, unsatisfactory for zoning on a regional scale. In order to carry this out effectively, the utilization of spatial estimation of the relative levels of landslide hazard in the various areas was considered opportune. These areas were classified according to their levels of proneness to landslide activity without taking recurrence periods into account. Various techniques were developed in order to obtain upheaval numerical estimates. The method used in this study, which was applied in the area of Potenza, is based on techniques derived from artificial intelligence (Artificial Neural Network—ANN). This method requires the definition of appropriate thematic layers, which parameterize the area under study. These are recognized by means of specific analyses in a functional relationship to the event itself. The parameters adopted are: slope gradient, slope aspect, topographical index, topographical shape, elevation, land use and lithology.     

Analysis of deformation fabric in an Alkaline Complex (Koraput): Implications for time relationship between emplacement,fabric development and regional tectonics

The Koraput Alkaline Complex (KAC) lies on the NE-SW trending Sileru Shear Zone (SSZ) separating the Proterozoic Eastern Ghats Province from the Archaean Indian craton. The core of the KAC is made of hornblende gabbro, which is rimmed by a band of nepheline syenite in the east and syenodiorite in the west. The timing of magmatism with respect to the SSZ is disputed. The KAC was deformed during emplacement, and a magmatic foliation related to the syn-emplacement deformation, D ₁ , is present in the gabbroic core. The dominant D ₂ -related field fabric strikes NESW and is penetrative in parts of the gabbro and marginal lithologies. E-W trending D ₃ shear zones cut across the complex. Distinct textural domains resulted from strain partitioning during deformation. Parts of the complex with magmatic textures constitute Domain-1, while D ₂ and D ₃ fabric zones comprise Domains-2 and 3 respectively. Temperatures in the KAC initially decreased following D ₁ , but increased through D ₂ and D ₃ . Anisotropy of magnetic susceptibility (AMS) studies show that the magnetic fabric generally follows S ₁ in Domain-1. While the magnetic fabric in Domain-2 is dominantly parallel to S ₂ , some of it parallels S ₁ . The latter is a relict D ₁ fabric that is recognized from AMS analysis but is obliterated in the field, which confirms that the KAC pre-dates the SSZ. The response of magnetic fabrics to temperature and implications of the study for Indo-Antarctica amalgamation are discussed.     

Composite magnetic fabric deciphered using heating treatment

In a number of AMS studies, the presence and deciphering of composite magnetic fabrics is of major importance for a correct interpretation of the data. On the basis of several examples from intrusive rocks (diorites and dolerites) we show that the use of laboratory heatings can help to extract at least one component of the composite magnetic fabrics usually present. The procedure includes comparison of the fabrics measured after stepwise laboratory heating with the fabrics determined by tensor difference and by linear regression analysis. In the diorite samples, the measured AMS results from the superimposition of different component fabrics and does not correspond exactly to any of these fabrics. In these dykes, isolated magnetic fabric during thermal treatment corresponds to that of the main magnetic mineral (Ti-poor titanomaghemite) and reveals an unknown structure. In volcanic flow or doleritic dykes, a “parasitic” fabric related to late or post-magmatic evolution superimposed to the flow fabric can produce important scattering of the AMS principal directions. Decomposition of magnetic fabric during thermal treatment allows isolation of the flow fabric.     

On the development of a magnetic susceptibility-based tracer for aeolian sediment transport research

Aeolian processes – the erosion, transport, and deposition of sediment by wind – play important geomorphological and ecological roles in drylands. These processes are known to impact the spatial patterns of soil, nutrients, plant-available water, and vegetation in many dryland ecosystems. Tracers, such as rare earth elements and stable isotopes have been successfully used to quantify the transport and redistribution of sediment by aeolian processes in these ecosystems. However, many of the existing tracer techniques are labor-intensive and cost-prohibitive, and hence simpler alternative approaches are needed to track aeolian redistribution of sediments. To address this methodological gap, we test the applicability of a novel metal tracer-based methodology for estimating post-fire aeolian sediment redistribution, using spatio-temporal measurements of low-field magnetic susceptibility (MS). We applied magnetic metal tracers on soil microsites beneath shrub vegetation in recently burned and in control treatments in a heterogeneous landscape in the Chihuahuan desert (New Mexico, USA). Our results indicate a spatially homogeneous distribution of the magnetic tracers on the landscape after post-burn wind erosion events. MS decreased after wind erosion events on the burned shrub microsites, indicating that these areas functioned as sediment sources following the wildfire, whereas they are known to be sediment sinks in the undisturbed (e.g. not recently burned) ecosystem. This experiment represents the first step toward the development of a cost-effective and non-destructive tracer-based approach to estimate the transport and redistribution of sediment by aeolian processes. © 2018 John Wiley & Sons, Ltd.     

MT探测居里面深度的可行性研究

居里面是地球内部铁磁性物质向顺磁性物质转换的界面,在温度略低于居里点时物质磁化率会快速升高,这被称为Hopkinsin峰.对于地球内部而言Hopkinsin峰是只有几百米至几公里厚的薄层,由于其与居里温度的关系,因此其底界面的深度可以作为居里面深度的估计.传统的居里面深度探测方法包括谱分析方法、等层模型方法和温度－深度剖面法.这些方法是人们研究地球内部热结构和居里面深度的重要手段,但是谱分析和等层模型的结果均有一些固有的缺点,如横向分辨率太低等;而地热方程的结果则受地表因素影响十分严重,并且地球内部的热源分布也不是十分清楚,这导致了其结果是不可靠的.本文提出用MT方法探测居里面深度,通过对几种简单一维模型进行的正反演数值试验,论证了该方法的可行性.结果表明,用MT方法研究居里面性质,不但可以得到居里面深度,还可以得到居里面顶部Hopkinsin峰所对应介质的电学和磁学性质,但必须同时反演岩层的电阻率和磁导率,才能获得较为可靠的居里面深度估计.     

Using magnetic susceptibility to assess soil degradation in the Eastern Rif,Morocco

The soil in the Rif, Morocco, is at serious risk because increasing anthropogenic pressures are gradually transforming large natural areas into farmland. The distribution of magnetic minerals within the soil profile can be used to assess soil development and degradation. The soils in the study area are severely eroded because of a combination of highly erodible soils, intense rainstorms and scarce vegetation cover. To sample of representative soil profiles, lithology, slope gradient and land use were considered. The ranges of magnetic susceptibility in the soil profiles distinguished between two primary soil groups. Magnetic susceptibility varied in the soil profile and along the soil toposequence, and the variations were related to the differences in the original magnetic composition and the influence of main erosion factors. Lithology is the main factor contributing to the variation in magnetic susceptibility. The magnetic susceptibility values in soils on Tertiary marls (χ = 13·5 × 10^?8 m³ kg^?1) differed significantly from those on Quaternary terraces (χ = 122·1 × 10^?8 m³ kg^?1). Slope affected the distribution of magnetic susceptibility because of the continuous loss of topsoil in some parts of the slope and the deposition of eroded soil in others. Elimination of the natural vegetation cover and a shift to cultivated land for cereals has had a negative impact on soil development and, on similar slopes and substrates, magnetic susceptibility decreased significantly in cultivated soils. The soils on steep slopes that had natural vegetation cover retained the magnetic minerals better than did those on gentler slopes that were under cultivation. Grazing, clearing and, especially, tilling has weakened the soil and made it much more vulnerable to erosion. An analysis of the main factors causing erosion will help to promote rational use of the land and to establish conservation strategies in such fragile agroecosystems. Copyright © 2009 John Wiley & Sons, Ltd.