基于DCT-PLS算法的MODIS LST缺值填补方法研究

受云层、传感器误差等因素影响,中分辨率成像光谱仪（MODIS）获取的地表温度产品（LST）在时间和空间上存在大量缺失数据,严重影响了基于时间序列数据的分析与应用。本文引进一种基于离散余弦变换与惩罚最小二乘的多维数据快速平滑方法（DCT-PLS）,利用数据集自身的时间和空间信息填补缺值。本文在粤港澳大湾区开展实证研究,将DCT-PLS算法用于填补该地区2001年1月—2017年12月的月值MODIS LST数据缺值,并引入人工模拟缺值对算法进行误差分析与精度验证。算法误差分析结果表明,填补误差主要来源于三维算法对数据集中有偏LST时间信息的使用,并因此产生显著高估或低估的填补结果。基于此,本文提出了利用MODIS LST数据集自身时间序列信息自动计算获取有效辅助LST信息的优化策略,从而实现填补算法计算效率和精度的提升：平均计算时间从12.0 s提高至1.7 s,平均R从0.94提高至0.97,平均RMSE从1.94 K提高至0.74 K（相较于三维算法）。在大湾区的填补结果表明（日间结果：R=0.98、RMSE=0.79 K;夜间结果：R=0.99、RMSE=0.56 K）,优化后的DCT-PLS算法可以快速鲁棒地填补MODIS LST月值数据产品中的缺值,并且具有稳定性强、不依赖外部数据集的计算特性,能够适应长时间序列MODIS LST缺值填补。     

改进的ELU卷积神经网络在SAR图像舰船检测中的应用

随着航天技术的发展,我国SAR载荷的探测体系呈现多种类、多分辨率的发展趋势。传统的检测识别方法很难适应多分辨率、多种类的SAR图像数据,从而需要寻求一种能从多分辨率的图像数据中提取有效特征的方法。智能化发展非常迅速,本文基于SAR图像的特点,提出了改进的ELU激活函数卷积神经网络的方法,建立了结合ELU激活函数和二次代价函数的深度学习模型。同时,在训练样本中建立样本特征与所在分类中心的距离函数,用模糊支持向量机（FSVM）对提取的特征进行了分类。试验结果表明,本文方法提高了SAR图像舰船检测的抗噪性,并且检测率达到了98.6%。     

Two New Species of Protocedroxylon Gothan (Pinaceae) from the Middle Jurassic of Eastern Inner Mongolia, NE China

Two new coniferous wood taxa, Protocedroxylon zhangii sp. nov. and P. zhalantunense sp. nov., are described from the Middle Jurassic Wanbao Formation in Zhalantun City, Inner Mongolia, NE China. The new discovery represents the first record of petrified wood in the Wanbao Formation, and contributes to further understanding the floral composition, especially that of the forest, of the Wanbao Formation. Anatomically, the two new species are characterized by having a mixed type of radial pitting and Abietineentüpfelung. In extant conifers, Abietineentüpfelung is considered to be characteristic of the wood of Pinaceae. The new discovery further indicates that the first occurrence of Abietineentüpfelung in conifer-like wood should be earlier than the Kimmeridgian. The palaeolatitudinal distribution pattern of Protocedroxylon indicates that the genus is a palaeobiogeographically consistent group, which was probably restricted to wetter and cooler temperate climates. In combination with distinct growth rings, the occurrence of Protocedroxylon suggests that a cool temperate and seasonal climate might have prevailed during the sedimentation period of the Wanbao Formation in eastern Inner Mongolia, NE China. Such a view is also in agreement with the palaeoclimatic conditions indicated by the plant megafossils of the Wanbao Formation.     

西澳Darling Range地区红土型铝土矿地球化学特征及源区研究

Darling Range位于西澳大利亚地区,红土型铝土矿资源丰富。本文对Darling Range地区红土型铝土矿及相关岩石开展了元素地球化学分析,并对铝土矿中的碎屑锆石和其下伏的花岗岩开展了LA-ICP-MS锆石U-Pb测年,目的是探讨西澳地区红土型铝土矿的成矿作用及其源区。结果表明,西澳地区红土型铝土矿的主量元素以Al2O3、Fe2O3、SiO2和TiO2为主,Al2O3与SiO2呈显著的负相关关系,与TiO2则呈现显著的正相关关系,表明铝土矿的形成过程是一个去Si富Al、Ti的过程。铝土矿微量元素富集Th、U、Nb、Ta、Zr、Hf、Ti,相对亏损Ba、K、Sr、P、Sm;稀土总量较低,为1.36×10~(-6)~65.58×10~(-6),稀土元素球粒陨石标准化分布曲线略微向右倾斜,富集轻稀土。铝土矿碎屑锆石U-Pb年龄分布于1167Ma和1267Ma与2539~2696Ma(16颗)两个年龄段,分析锆石来源前者可能来自Albany-Fraser造山带;后者加权平均年龄为2579±15Ma(MSWD=1.16),与研究区本次获得的太古宙花岗岩加权平均年龄2585±12Ma(MSWD=0.83)在误差范围内一致,表明铝土矿中的锆石除了来自本区太古宙花岗岩,还有少量的锆石来自Albany-Fraser造山带。结合元素地球化学钛率(Al2O3/TiO2)、lgCr-lgNi、微量元素比值、稀土元素配分模式对铝土矿源区进行示踪表明,Darling Range地区的太古宙花岗岩为红土型铝土矿的主要物质来源。     

土柱变密度溶质运移试验推求水平及垂直方向弥散度研究

基于溶质运移对流弥散理论的变密度海水入侵模型广泛用于海水入侵研究,而水动力弥散系数是影响模型模拟效果的关键性参量之一。利用传统土柱溶质运移试验结合旁侧抽水,采用数值反演法成功获取了水平及垂直两方向弥散度。相对于传统方法,在不增加试验复杂度的前提下,同时推求了不同方向的弥散度,提高了试验效率,节约了试验成本,可广泛用于测定水动力弥散系数等参数。     

降雨条件下边坡有限元强度折减法计算平台开发及其应用

降雨是导致边坡失稳的主要因素。目前,使用有限元强度折减法进行降雨滑坡计算时,过程比较复杂,计算工作量大。同时,对降雨边界的处理更加大了计算的复杂性,从而影响了强度折减法在降雨滑坡计算中的推广应用。为了更加准确、高效地将有限元强度折减法应用于降雨滑坡计算,首先基于Geostudio软件采用迭代算法对降雨入渗边界进行处理,通过计算得到每一时刻边坡内的水压分布,然后利用Fortran编制程序将水压转化为等效结点力并导入ABAQUS中,之后结合“基于场变量的有限元强度折减法”对ABAQUS进行二次开发,最终建立降雨条件下边坡有限元强度折减法计算平台。该平台利用Geostudio考虑了降雨入渗边界的复杂性,克服了ABAQUS软件中不能准确模拟降雨边界的问题。同时采用ABAQUS软件提供的场变量,对有限元强度折减法的计算过程进行简化,提高了计算效率,经过一次计算即可求得每一降雨时刻的边坡稳定系数。通过与传统极限平衡法进行比较,发现两者得出的结果具有较好的一致性,表明该计算平台是稳定、可靠的。最后采用该平台对降雨滑坡机理进行研究,研究结果可为边坡防减灾提供参考。     

空气潜孔锤钻进典型事故原因分析及处理措施

贵州省地质矿产勘查开发局就“贵州复杂地层深井钻探工艺应用研究”科研项目中“空气潜孔锤钻进技术” 部分进行了试验研究,在试验过程中曾出现了锤头断裂、硬质合金齿折断、冲击器堵塞并阻卡、泥环包裹、卡埋钻、井眼弯曲等典型事故。详细分析了出现这些典型事故的原因,形成了避免和解决这些典型事故的系统技术措施。该成果经现场检验效果良好。本文对该成果进行了全面阐述,以期为同类施工提供借鉴。     

Multiscale modeling and analysis of compaction bands in high-porosity sandstones

We present a multiscale investigation on the initiation and development of compaction bands in high-porosity sandstones based on an innovative hierarchical multiscale approach. This approach couples the finite element method and the discrete element method (DEM) to offer direct, rigorous linking of the microscopic origins and mechanisms with complex macroscopic phenomena observed in granular rocks such as strain localization and failure. To simulate compaction band in granular cementitious sandstone, we adopt a bonded contact model with normal and tangential interparticle cohesions in the DEM and propose a dual-porosity structure consisting of macro-pores and interstitial voids for the representative volume element to mimic the typical meso-structure of high-porosity sandstones. In the absence of particle crushing, our multiscale analyses identify debonding and pore collapses as two major contributors to the formation of compaction bands. The critical pressures predicted by our simulations, corresponding to surges of debonding and pore collapse events, agree well with the estimations from field data. The occurrence patterns of compaction band are found closely related to specimen heterogeneity, porosity and confining pressure. Other deformation band patterns, including shear-enhanced compaction bands and compactive shear bands, were also observed under relatively low confining pressure conditions with a rough threshold at \(0.55P^{*}\) (\(P^{*}\) is the critical pressure) on the failure envelop. Key microscopic characteristics attributable to the occurrence of these various deformation patterns, including fabric anisotropy, particle rotation, debonding and pore collapse, are examined. Shear-enhanced compaction bands and pure compaction bands bear many similarities in terms of these microscopic characteristics, whereas both differ substantially from compactive shear bands.