高光谱影像信息向量机分类

信息向量机是一种基于贝叶斯理论的稀疏高斯过程方法,其模型训练速度快、内存耗费小、稀疏性强,具有良好的预测性能。本文从高斯过程回归模型出发,提出了一种基于信息向量机的高光谱影像分类方法,针对高斯过程分类中的非高斯噪声模型,采用假定概率滤波算法将分类问题转化为回归问题,通过最大化边缘似然函数进行模型训练,选择活动子集中的信息向量数量达到了稀疏的目的。通过ROSIS影像试验,表明了基于信息向量机的高光谱影像分类方法的优势。     

湘东锡田地区太和仙金多金属矿找矿潜力分析

文章通过对太和仙金多金属矿区地层、构造、岩浆岩、矿体特征等的综合研究与分析,认为该矿区矿源层分布广泛,成矿构造条件优越,成矿序列明显,矿床属岩浆热液成因的破碎带蚀变岩型金多金属矿床,并以此为基础,建立了太和仙矿区破碎带蚀变岩型金多金属矿的成矿模式;重力异常特征、水系沉积物异常特征显示矿区处于隆起的热液活动区,下伏可能存在一隐伏岩体。推测太和仙矿区找矿潜力大,可作为寻找中大型以上规模金多金属矿床的有利地段。     

A Spatial Consistency Quality Control Method for Daily Surface Temperature Observations

A new spatial consistency quality control method (SRF) based on the spatial regression test (SRT) and random forest (RF) was adapted to identify potential outliers in daily surface temperature observations in this article. For the new method, the SRT method was used to filter the data and the RF method was used to conduct regression. To evaluate the performance of the quality control method, the SRF, SRT and RF methods were applied to a surface temperature dataset with seeded errors from different regions of China from 2005 to 2014. Compared to SRT and RF, the results indicate that the SRF method outperforms the other two methods for the most cases. And the results of the comparison led to the recommendation that the SRF method improves the regression accuracy of traditional spatial consistency quality control methods and reduces the runtime of random forest through data refinement.     

1–100 μm energy spectra and 2.3–4 μm spectrophotometry of MSC stars

The JHKL' photometry and 2.3–4 μm spectrophotometry of some M giants, S type stars and carbon stars are presented in this paper. It is found that in combination with IRAS data, the energy spectra in 1–100 μm of S type stars are intermediate between those of M giants and carbon stars, which are obviously different. The spectrophotometry in the near infrared shows that, besides carbon stars, which have HCN and C₂H₂ strong absorptions at 3.1 μm, some S type stars have the similar but weaker absorption in the same spectral region. However, no trace of any absorption at 3.1 μm can be seen in M giants. These results probably provide more evidence for the M-S-C sequence in the late stage of stellar evolution.     

Analysis on the interplanetary causes of the great magnetic storms in solar maximum (2000-2001)

In this paper, we analyze the interplanetary causes of eight great geomagnetic storms during the solar maximum (2000-2001). The result shows that the interplanetary causes were the intense southward magnetic field and the notable characteristic among the causal mechanism is compression. Six of eight great geomagnetic storms were associated with the compression of southward magnetic field, which can be classified into (1) the compression between ICMEs (2) the compression between ICMEs and interplanetary medium. It suggests that the compressed magnetic field would be more geoeffective. At the same time, we also find that half of all great storms were related to successive halo CMEs, most of which originated from the same active region. The interactions between successive halo CMEs usually can lead to greater geoeffectiveness by enhancing their southward field B_s interval either in the sheath region of the ejecta or within magnetic clouds (MCs). The types of them included: the compression between the fast speed transient flow and the slow speed background flow, the multiple MCs, besides shock compression. Further, the linear fit of the D_st versus gives the weights of and Δt as α=2.51 and β=0.75, respectively. This may suggest that the compression mechanism, with associated intense B_s, rather than duration, is the main factor in causing a great geomagnetic storm.     

The crystal structure of (Fe<Subscript>4</Subscript>Cr<Subscript>4</Subscript>Ni)<Subscript>9</Subscript>C<Subscript>4</Subscript>

(Fe4Cr4Ni)9C4 is a metal carbide mineral formed by combination of Fe, Cr and Ni with C. It occurs in a chromite deposit in the Luobusha ophiolite, Tibet. Based on the determination of its crystal structure, the empirical formula is (Fe4.12Cr3.84Ni0.96)8.92C3.70 and the simplified formula is (Fe4Cr4Ni)4C9. The mineral is hexagonal with a = 1.38392(2) nm, c = 0.44690(9) nm, pace group P63 m c, Z=6 and the calculated specific gravity Dx = 7.089 g/cm3. Fe, Cr and Ni occupy different crystallographic sites and their coordination numbers are approximately 12, forming an alternate stacking sequence of flat and puckered layers along the c axis. Some metallic atoms have a defect structure. The interatomic distances of Fe, Cr and Ni are 0.2525-0.2666 nm, and the distances between Fe, Cr, Ni and C are 0.1893-0.2169 nm. The coordination number of carbon is 6. It occurs in interstices of the metallic atoms Fe, Cr and Ni to form trigonalprismatically coordinated polyhedra. These coordination polyhedra are linked with each other via shared corners or shared edges into a new type of metal carbide structure.     

半参数测量平差模型与系统误差处理

考虑半参数测量模型L=Bx+S+Δ,x$Rd为未知回归参数,S为未知Borel函数。本文首先利用自然样条函数法,找到符合条件的非参数自然插值样条函数。其次利用偏残差法并综合最小二乘法,导出了这种平差方法的解算公式。最后,用一个算例说明了此方法的优越性。     

当前国内酸雨研究进展

回顾了我国在酸雨研究方面所获得的成果,包括酸雨的时空分布特征、影响因素、生态影响模拟以及控制对策等4个方面的内容。大量的研究表明,近些年来酸雨污染的范围有所增加、程度有所加重。酸雨的形成是多种自然和人为因素共同作用的结果,与特定的气象条件、天气类型和大气环流等有密切的关系。针对酸雨污染有所加重的趋势,必须采取有效的控制对策。     

CERES-Wheat模型在我国小麦区的应用效果及误差来源

气候模型与作物模型耦合是评价未来气候变化对作物生产影响的常用方法之一, 但当两者结合时, 存在着空间和时间尺度差异问题, 将作物模型升尺度到区域是解决该差异的一种方法。将CERES-Wheat模型升尺度进行区域模拟, 利用区域校准后的CERES-Wheat模型, 模拟了1981—2000年全国各网格小麦产量, 与同期农调队调查产量相比较, 以探讨CERES-Wheat模型在我国小麦区的模拟效果及误差来源。结果表明:全国小麦产量的区域模拟值与农调队调查产量的相对均方根误差为27.9%, 符合度为0.75, 全国59.2%的模拟网格相对均方根误差在30%以内, 其中相对均方根误差小于15%的占26.3%;各区的效果不同, 种植面积最大的小麦种植生态2区, 模拟效果最好。总体来说, CERES-Wheat的区域模拟, 可以反映产量变化规律, 能为宏观决策提供相应信息, 尤其是在主产区; 但区域模拟中还存在一系列误差, 今后还需进一步研究。