大数据环境下内蒙古浩布高地区铅锌多金属矿智能矿产预测研究

传统机器学习算法已广泛应用于矿产预测,但面对地质大数据的高维稀疏、不平衡小样本等特性仍缺乏有效处理和分析的方法,设计适合地质大数据特点的机器学习算法是智能矿产预测亟需解决的新问题。本文以内蒙古浩布高地区的铅锌多金属矿产预测为例,提出了一种面向地质大数据的半监督协同训练矿产预测模型。首先对研究区地质找矿信息和地球化学异常信息进行定量分析,提取断裂构造、二叠系地层、燕山期侵入岩、地层与岩体接触带、围岩蚀变及Pb、Zn、Sn、Cu地球化学异常共9种找矿因子。然后利用递归特征消除法优选找矿因子组合,不包括Sn异常在内的8个找矿因子组合被选为最优组合。最后,利用支持向量机和随机森林算法作为基分类器进行半监督协同训练矿产预测,绘制成矿概率分布图。ROC曲线和预测度曲线分析结果表明,半监督协同训练模型的AUC值和预测效率都高于随机森林和支持向量机模型。研究结果也为大数据环境下的智能矿产预测提供了一种新的思路。     

The Derivation and Validation of TBM Disc Cutter Wear Prediction Model

The problem of disc cutter wear is inevitable when shield or TBM excavating hard rock for a long distance, thus, the study of disc cutter wear model has an important project value on predicting its service life and replacement opportunity. It is put forward by analyzing disc cutter wear mechanism that the main wear form is abrasive wear, which is based on plastic removal mechanism. Then, disc cutter wear rate and linear wear rate prediction models are obtained by approximate calculation and mathematical deduction, which are based on Rabinowicz equation and CSM model. At last, the two models are verified through field test data from three projects, and the results show that the prediction model can accurately reflect the real wear situation of disc cutter.     

Progress and Prospect of Statistics and Assessment of Large-scale Natural Disaster Damage and Losses

In June 2014, the "Statistics System for the Damage and Loss of Large-scale Natural Disasters" (SSDLLND) was issued by the Ministry of Civil Affairs and the Office of National Disaster Reduction Committee, which marked that the statistics and assessment of China's catastrophic natural disaster damage and losses formally entered a new stage of institutionalization. On the basis of analyzing the five major international disaster damage and loss assessment systems, including HAZUS-MH, ECLAC, DaLA, EMA-DLA and PDNA, the differences between the “SSDLLND” of China and five major international systems were compared from the statistics and assessment contents and indicators. Combined with the statistics and assessment practices of China’s large-scale disaster damage and losses and the characteristics of international systems in recent years, the future development of the SSDLLND were proposed in three aspects: Enriching and improving the framework of damage and loss statistics content, stepwise improvement of disaster impact assessment methods (such as the ecological capital loss assessment, tourism industry loss assessment due to the disasters, etc.), and improving indicators and parameters of loss statistics. The study has an important practical significance for improving the statistics and assessment system of the damage and loss of major natural disasters and better serving the post-disaster recovery and reconstruction decision-making.     

Intraseasonal oscillation of the rainfall variability over Rwanda and evaluation of its subseasonal forecasting skill

非洲中东部地区的经济主要依靠自给农业支撑,该地区农业经济对降水的变化尤为敏感.本文以卢旺达为例,观测分析指出卢旺达的次季节降雨主要集中在10-25天;根据次季节尺度降水变率的单点相关方法,发现卢旺达的次季节降水变率和周围区域变化一致;进一步合成结果显示该地区次季节降水变率与异常西风有关,这可追溯到赤道地区西传的赤道Rossby波.最后,本文评估了当前动力模式ECMWF对 卢旺达地区(即非洲中东部)次季节降水变率的预报能力,发现EC模式在对该区域降水和相关风场指数的预报技巧都在18天左右,且预报技巧表现出一定的年际差异,这可能与热带太平洋的背景海温信号有关.该工作增进了当 前对非洲中东部地区的次季节降水变率和预测水平的认知,并且对该地区国家粮食安全和防灾减灾具有启示性意义.     

Intraseasonal oscillation of the southwest monsoon over Sri Lanka and evaluation of its subseasonal forecast skill

斯里兰卡的雨季发生于5-9月间,主要受西南季风的控制.本文发现该地区的西南季风降水存在很强的次季节变率,主导周期为10-35天.降水的季节内变化与西传的异常气旋有关.进一步,利用S2S比较计划中欧洲中心的数值预报模式(ECMWF)提供的回报试验数据,评估了当今动力模式对斯里兰卡西南季风次季节变化的预报技巧.结果显示,对季风指数的预测技巧超过30天,而对降水指数的预测技巧大约两周,且模式的预报技巧具有明显的年际差异.分析表明,能否正确模拟出大尺度环流对热带对流的响应是影响斯里兰卡降水预测的重要因子.     

An enhanced approach for surface flow routing over drainage‐constrained triangulated irregular networks

The accuracy and efficiency of the simulations in distributed hydrological models must depend on the proper estimation of flow directions and paths. Numerous studies have been carried out to delineate the drainage patterns based on gridded digital elevation models (DEMs). The triangulated irregular network (TIN) has been increasingly applied in hydrological applications due to the advantages of high storage efficiency and multi‐scale adaptive performance. Much of the previous literature focuses mainly on filling the depressions on gridded DEMs rather than treating the special cases in TIN structures, which has hampered its applications to hydrological models. This study proposes a triangulation‐based solution for the removal of flat areas and pits to enhance the simulation of flow routing on triangulated facet networks. Based on the drainage‐constrained TIN generated from only a gridded DEM by the compound point extraction (CPE) method, the inconsistent situations including flat triangles, V‐shape flat edges and sink nodes are respectively identified and rectified. The optimization algorithm is an iterative process of TIN reconstruction, in which the flat areas are generalized into their center points and the pits are rectified by embedding break lines. To verify the proposed algorithm and investigate the potential for flow routing, flow paths of steepest descent are derived by the vector‐based tracking algorithm based on the optimized TIN. A case study of TIN optimization and flow path tracking was performed on a real‐world DEM. The outcomes indicate that the proposed approach can effectively solve the problem of inconsistencies without a significant loss in accuracy of the terrain model.     

内蒙古平安地元古代构造混杂岩的发现与地质意义

正研究区位于内蒙古东部天山镇西平安地地区,古生代大地构造位置由南向北可划分为华北板块、白乃庙岛弧和索伦—林西缝合带(图1)。Xiao等(2015)和Jian等(2012)等认为温都尔庙—柯丹山—西拉木伦蛇绿混杂岩带是华北板块和西伯利亚板块的缝合带,研究区则位于该混杂岩带以南白乃庙岛弧一侧。     

青海茫崖石棉矿区变质超基性岩的蚀变演化序列：岩石化学及矿物学证据

青海茫崖石棉矿区超基性岩体是由原岩以纯橄岩、辉橄岩和橄辉岩为主体组成的富镁质超基性岩体,经历自变质和后期多期热液的叠加变质蚀变作用,经蛇纹石化后形成蚀变完全的蛇纹岩岩体,其中部分蛇纹岩又进一步发生滑石化及碳酸盐化蚀变为滑石菱镁片岩、菱镁滑石片岩、滑石片岩和菱镁岩等。本文在野外地质调查基础上,在室内通过镜下岩矿综合鉴定、全岩化学成分分析以及电子探针成分分析等手段进行了岩石化学特征、矿物学特征及其蚀变演化过程研究。结果表明,该变质超基性岩体蛇纹岩主要特征组合矿物为蛇纹石(利蛇纹石、叶蛇纹石、纤蛇纹石)、磁铁矿、菱镁矿、滑石、水镁石、铬铁矿,变余矿物有斜方辉石、单斜辉石和铬铁矿,滑石菱镁片岩类主要组成矿物为菱镁矿、滑石、蛇纹石及磁铁矿,局部可见石英脉。该地区变质超基性岩体较完整地记录了橄榄岩水化、滑石化及碳酸盐化作用过程的各个阶段,超基性岩蚀变演化过程主要有两个作用阶段:(Ⅰ)橄榄石、辉石类矿物的蛇纹石化作用及蛇纹石绿泥石化作用;(Ⅱ)富Ca、CO2流体交代蛇纹石、滑石及水镁石的碳酸盐化作用。蛇纹石化等变质蚀变作用促进了Si、Mg及Fe元素化学活动性,使元素发生富集与迁移,对于次生矿物的形成与演化起到了一定的催化作用。多期不同组成流体热液的交代作用过程,清晰地展示了利蛇纹石、纤蛇纹石和叶蛇纹石的演化序列,以及滑石、水镁石、铬铁矿和磁铁矿的形成过程及标形特征。     

华北克拉通东缘金岭杂岩体岩浆源区及构造背景:来自岩相学、岩石地球化学及年代学的证据

金岭杂岩体由细粒黑云角闪闪长岩、中细粒黑云角闪闪长岩、中细粒黑云角闪二长闪长岩和细粒角闪二长闪长岩组成,是鲁西地区典型矽卡岩型富铁矿(金岭铁矿)控矿岩体。本次研究对细粒黑云角闪闪长岩和细粒角闪二长闪长岩进行了锆石LA‐ICP‐MS U‐Pb定年,其结果分别为129.2±3.2 Ma和132.8±1.2 Ma,表明该杂岩体的侵位时代为早白垩世。样品SiO2、K2O和Na2O含量分别介于54.17%~63.73%、1.92%~4.76%和3.10%~5.41%之间,K2O/Na2O为0.58~0.94,A/CNK为0.60~0.93,具有轻稀土富集,重稀土亏损的右倾型稀土配分模式,轻、重稀土分馏程度中等((La/Yb)N=9.94~23.49),Eu异常不明显(δEu=0.84~1.10),具中—弱负Ce异常(δCe=0.56~0.92)。样品以富集大离子亲石元素(Ba、K、U、Pb等)、亏损高场强元素(Nb、Ta、Ti)以及高Sr/Y为特征。金岭杂岩体为燕山晚期岩浆活动产物,属准铝质高钾钙碱性系列,岩浆主要来源于富集岩石圈地幔的部分熔融,并在岩浆上升侵位的过程中有地壳物质的同化混染。燕山晚期华北克拉通在古太平洋板块俯冲后后撤引起的板内伸展环境下,增厚陆壳减薄阶段,岩浆上侵就位形成金岭杂岩体。