基于深度表征学习的紫外极光卵图像聚类

极光受太阳风驱动的地磁亚暴等大尺度动力学影响，其形态及演化因不同的太阳风-磁层-电离层耦合作用可能表现不同。目前，极光卵及其形态的归类大多依据极光演化理论作主观定性分析，没有明确的分类标准，故难以借助统计分析方法和有监督分类模型开展客观定量研究。建立了基于深度表征学习的紫外极光卵图像聚类模型（MoCo-GMM），并利用空间环境参数设计了评估模型物理合理性的方法，在大规模POLAR卫星紫外极光卵图像数据上进行了实验，聚类结果不仅具有良好的簇内凝聚性和簇间分散性，且具备一定的物理可解释性，有效实现了基于图像的极光卵及其形态的客观归类。     

基于紫外极光图像的极光电集流指数模型

极光电集流指数AE是描述地磁亚暴强弱的重要指标,且与极区磁层扰动及极光粒子沉降过程密切相关。因此,建立更加准确的极光电集流指数模型对空间天气的研究具有重要意义。利用1997年POLAR卫星紫外极光图像数据探究了紫外极光图像中极光强度IAP的空间分布与AE指数在不同季节的相关性,并在此基础上提出了基于紫外极光图像的AE指数模型。以网格化方法提取极光强度空间分布特征,采用广义回归神经网络,通过相关系数法和方差选择法构建Cor-GRNN和Var-GRNN两种AE指数模型,并针对冬至月份、夏至月份、分点月份3个季节分别进行训练。研究结果表明,AE指数与IAP具有相似的半年变化趋势,其相关性在不同季节差异较大。相比于太阳风驱动下的AE指数神经网络预测模型,基于极光图像的AE指数模型在ERMS和R2标准上均优于其他模型,其中归一化ERMS小于0.1,模型对于AE指数变化的可解释度提升了10%左右。     

喉区极光的机器识别

喉区极光是一种发生在电离层对流喉区附近的极光现象，是极光卵向低纬侧延伸出的南北向分立结构，其可能对应由磁鞘高速流与磁层顶作用引发的磁层顶重联过程.喉区极光研究对深入理解太阳风—磁层—电离层耦合过程具有重要意义.从长期观测所积累的大量全天空极光观测数据中准确高效识别出喉区极光结构，是开展喉区极光统计研究的基础.本文利用北极黄河站2003—2017年全天空成像仪的极光观测数据，建立了喉区极光图像标注数据集；基于密集连接卷积神经网络（DenseNet）对极光图像全局高维表征的自动学习，首次实现了喉区极光结构的机器识别.算法对喉区极光识别准确率达96%，且具有良好的泛化性能.研究表明基于深度学习的图像识别方法可用于从海量极光观测数据中自动识别喉区极光事件.      

极向运动极光结构和喉区极光光学观测特征的对比分析

利用中国北极黄河站高时间分辨率的三波段全天空成像仪极光观测数据，联合太阳风和行星际磁场等观测，分析了极向运动极光结构（PMAFs）和喉区极光的形成及演化特征.研究发现:一系列PMAFs与喉区极光事件同时出现在观测视野中，其中PMAFs主要发生在日侧极隙区极光卵赤道向边界的极向一侧，沿东西方向分布，点亮后向高纬运动；喉区极光紧靠PMAF一侧发生，从极光卵赤道向边界向低纬延伸，沿南北方向分布，点亮后向高纬偏西方向运动；观测期间PMAFs发生频率高于喉区极光；当PMAFs与喉区极光同时出现时，PMAFs可以与喉区极光几乎同时出现或略晚于喉区极光出现，持续时间较喉区极光短.观测结果表明:与PMAF相对应的磁层顶重联过程和与喉区极光对应的磁层顶凹陷导致的磁重联过程在日侧磁层顶上的相邻区域分别发生，两种极光事件的形成过程相对独立，可能不存在相互触发关系.      

地上天上看极光

极光是发生在地球两极周围高层大气中的一种十分宏伟壮丽的发光现象,也是唯一能用肉眼看见的地球高层大气中的一种物理过程。极光的形态变化万千,颜色绚丽多彩。极光的每一次出现,都好似大自然恩赐给人类的一幅美丽画卷,可惜地球上绝大多数人都生活在中低纬度地区,他们一辈子也没有欣赏这空中美景的机缘。     

极光西向涌浪的二维模式(线性部分)

极光西向涌浪(WTS)被定义为磁层亚暴开始的重要标志, 它是伴随带电粒子沉降过程在极光带电离层出现的特定的极光运动形态.本文给出了极光西向涌浪的二维动力学模式, 亚暴开始, 伴随粒子沉降在电离层产生离子密度梯度(电导率梯度), 在背景电场中激发出低频漂移波, 它的传播给出极光西向涌浪运动的主要特性.      

极光区电离层特性和磁层—电离层耦合

本文用EISCAT雷达的综合资料,分析了磁静条件下极光区电离层的季节、昼夜变化,与磁扰时的特殊形态进行了比较,讨论了磁层-电离层耦合对电离层形态的影响.     

极光电激流中的暂态过程

本义分析了极光电激流建立阶段的物理过程，认为在此期间存在着一个正反馈，从而导致一种暂态过程的产生。使用一种简化的物理模型，模拟了极光电激流在建立阶段的物理过程，计算结果表明，极光电激流中的暂态过程确实存在，电流、净电荷和电导率在治极光电激流的方向上均存在着传播的阶跃结构.计算结果与有关观测资料符合得很好.     

地球自转对北极黄河站观测日侧极光弧运动的影响

利用北极黄河站全天空极光数据,采用AACGM模型,将日侧极光弧映射到地磁坐标系,定量计算地球自转导致的极光弧运动速度.对于任意一条极光弧,其偏斜角定义为极光弧方向与当地地磁东西方向的夹角.研究发现,地球自转产生的速度由极光弧离开天顶的距离和偏斜角决定,其中偏斜角的影响更为重要,其还决定速度的方向.在4年的观测数据中,提取超过40000张出现极光弧的图像,计算极光弧偏斜角.计算结果表明,日侧极光弧的偏斜角随磁地方时增大而逐渐减小,并在大约10:00MLT（磁地方时）附近发生反转.由于偏斜角的反转,地球自转产生的极光弧运动在晨侧多为极向运动,到午后多为赤道向运动.相比午前,午后的运动更为明显,最大速度可超过300m·-1.      

北极黄河站极光亚暴期间低热层大气中性风研究

极光亚暴是地球空间基本的能量输入、耦合及耗散过程,其对低热层大气中性风的影响不容忽视,对这一问题进行深入研究具有重要意义.2010年11月,中国北极黄河站(78.92°N,11.93°E)安装了一台自主研制的全天空法布里-珀罗干涉仪(FPI)并开始进行正常观测,为中国首次获得了极光亚暴期间的FPI观测数据.根据2012年越冬期间的观测情况,对2012年11月12-14日及12月09-11日两个极光亚暴事件期间的数据进行处理,计算得到了5级干涉环对应的风场.亚暴期间风场变化与地磁活动变化的对比分析表明,风速的剧烈变化可能是由地磁活动剧烈扰动造成的.针对2012年11月13日00:00UT-02:00UT和2012年12月10日05:00UT-07:30UT的亚暴事件,将全天空相机拍摄到的极光图像与FPI干涉图像对应的视线风场进行对比分析.结果表明在极光活动中,视线风速加强的方向与极光弧的方向垂直,而在极光弧的平行方向,风速相对较小.      

Probing afternoon detached aurora and high-latitude trough based on DMSP observations

The present work displays the observations of an afternoon detached aurora along with ionospheric high-latitude trough. The event was observed by DMSP F17 on 19 September 2014. The afternoon detached aurora was isolated from the auroral oval and was located between 12:00–18:00 magnetic local time (MLT) and 65–70° geomagnetic latitude (MLAT). Particle observations indicate that the afternoon detached aurora was produced by energetic ring current ions with energies above ~10 keV where the main ion energy was likely to be above the upper limit of DMSP measurement (~30 keV). Magnetometer observation from the ground implies that the energetic ions were likely scattered by EMIC waves. Both the detached aurora and the auroral oval are found to be well inside the high-latitude trough with MLAT between ~64° and ~76° (68–80° GLAT). The auroral oval corresponds to a westward (sunward) plasma drift. It is expected that the westward drift transports the low-density plasma in the nightside toward the dayside, leading to the high-latitude trough formation. The afternoon detached aurora was well equatorward of the high-latitude trough, and the corresponding plasma drift was nearly zero. The plasma associated with the detached aurora is expected to be stagnant, and broaden the high-latitude trough equatorward.     

Images of the Earth's aurora and geocorona from the dynamics explorer mission

Several results from analyses of auroral and geocoronal images from the Dynamics Explorer Mission are summarized. (1) The motion of the transpolar arc of a theta aurora is found to be correlated with the y-component of the interplanetary magnetic field. The arc motion is in the general direction of the y-component. (2) A sequence of global images of a small auroral substorm shows the initial development of intense luminosities in a relatively small spatial region, or ‘bright spot’, in the pre-midnight sector of the auroral oval and a subsequent appearance of an expanding area of lesser intensities at lower latitudes and contiguous to the midnight boundary of the bright spot. This evolution of auroral luminosities is interpreted in terms of acceleration of electrons in the boundary layer of the magnetotail plasma sheet to produce the bright spot and subsequent injection into, and eastward drift within, the plasma sheet to form the diffuse area of lesser intensities. (3) A series of images of the Earth's geocorona in scattered solar Ly α emissions is used to obtain a best-fit spherical model of atomic hydrogen densities in the Earth's exosphere. A Chamberlain model provides an adequate fit to radial distances of 4.5 R_E, beyond which an exponential fit is used. The geocoronal tail is detected as an asymmetric increase in scattered Ly α intensities in the anti-solar direction.     

Saw-tooth substorms: Inconsistency of repetitive bay-like magnetic disturbances with behavior of aurora

The relationships between the magnetic disturbance onsets, aurora dynamics and particles injections at the geostationary orbit have been analyzed in detail for 25 sawtooth substorms. It is shown that inconsistency between the above signatures of the substorms onset is typical of the powerful sawtooth substorms, unlike the isolated (“classical”) magnetospheric substorms. The distinguishing feature of the aurora in case of saw-tooth substorms is permanently high level of auroral activity irrespective of the magnetic disturbance onsets and the double oval structure of the aurora display. The close relationship between the aurora behavior and the particle injections at geostationary orbit is also broken. The conclusion is made, that the classical concept of the substorm development, put forward by Akasofu (1964) for isolated substorms, is not workable in cases of the sawtooth disturbances, when the powerful solar wind energy pumping into the magnetosphere provides a permanent powerful aurora particle precipitation into the auroral zone.     

Advances in auroral imaging from space

Space imaging of the aurora provides unique quantitative information on the instantaneous global distribution and pattern of the aurora - a direct result of particle acceleration and precipitation processes occurring throughout the magnetosphere. Techniques for imaging from space have usually involved systematically scanning a detector across the earth using some combination of satellite motion and spin, scanning mirrors, and/or electronic scanning. The recent advent of CCD image sensor arrays opens up possibilities for vastly improved sensitivity and resolution in satellite imagers, although scanning systems retain significant advantages in situations where scattered light is a problem and where narrow band optical interference filters are needed.A new type of imaging instrument being built in Canada for the Swedish Viking satellite will employ a two-dimensional CCD array detector which is fibre-optically coupled to a micro-channel plate intensifier. It will produce images of low light level auroral emissions at the rate of one image per spin of the satellite (20 seconds). The instrument operates in the vacuum ultraviolet and uses a fast inverse Cassegrain optical system. Charge is shifted in the CCD imaging array in synchronism with the moving image so that exposure times of approximately one second will be achieved.     

午后极光强度与太阳风-磁层耦合函数的相关

利用1997年和1998年南极中山站多通道扫描光度计的地面观测数据和Wind卫星在弓激波上游对行星际磁场和太阳风参数的观测数据,对午后高纬极光强度与太阳风-磁层耦合函数之间的相关性进行定量研究.研究表明,午后630.0nm极光强度与太阳风-磁层耦合函数间有很好的相关,而557.7nm的相关性差一些;在考察的所有耦合函数中,午后极光受太阳风电场和能量的影响更直接;同时,行星际磁场的时钟角对午后极光也有很强的控制作用.      

Space Telescope observations of aurorae on the giant planets

For the distant giant planets, Uranus and Neptune, the observation of aurorae may be the best astronomical technique for the detection of planetary magnetic fields, with implications for the structure and composition of their interiors. Aurorae may be detected by emssion of H I Ly α (1216 Å) and by H₂ bands near 1600 Å. The latter are important for very faint aurorae because there is essentially no planetary, interplanetary or geocoronal scattering of sunlight to contaminate the signal. For Uranus, present IUE results suggest the presence of a strong aurora at Ly α, but the background and instrument noise levels are very high compared to the apparent signal. At 1600 Å, the IUE instrument noise renders the H₂ emission bands on Uranus marginal at best. No aurora has yet been observed on Neptune. For Jupiter, where the existence and general characteristics of the magnetic field are well established, there is disagreement between ground-based infrared and space-borne ultraviolet observations of the location of the aurorae. For all four giant planets, Space Telescope can improve upon the quality of current optical observations. For spectroscopy, the low resolution mode of the High Resolution Spectrograph (HRS) is particularly well suited to auroral observations because of its spectral range, adequate resolution and high sensitivity. For ultraviolet imaging through appropriate filters, the ST spatial resolution, expected to be of order 5 hundredths of an arc second, is also well suited to determine the spatial properties of the aurorae.     

一种地震动信号特征提取与分类方法

为了实时检测、识别和预警对地下基础设施的挖掘破坏活动，本文提出一种地震动信号特征提取与分类方法。通过提取小波包变换域和集合经验模态变换域的多域能量联合分布特征向量，构建改进的径向基神经网络分类模型，利用机器学习的方法提取稳定的信号多域融合特征，并实现准确的信号特征分类预测。由多类别挖掘信号的仿真实验结果可以看出，本文的算法和模型能有效提升地震动信号分类的准确率，对地震动干扰信号具有较强的鲁棒性。