利用BP神经网络预测AE指数

采用的预报模式是一种全连接的BP网络模型,利用太阳风及行星际磁场的观测数据预报AE指数.神经网络输入选用ACE卫星数据,取5 min平均值,通过比较,选用4个预报参量.构造了预报参量时续为20 min,40 min和60 min依次递增的三个网络,分别进行训练和预测,并对行星际参量对AE指数影响的时续性进行了探讨.预报结果表明,全连接BP神经网络在AE指数的短期预报中是比较有效的,同时还提出了需要进一步改进的环节.      

行星际电场与Dst指数

利用ACE卫星的太阳风及行星际磁场观测数据和相应时期的Dst指数，分析了行星际电场的Dst指数的相关关系，讨论了行星际电场作为研究磁层和太阳风相互作用的良好参数的物理机制。结果表明：行星际电场与Dst指数有很好的相关性，并且在强和中等地磁活动基间，存在显著的突变特征曲线；相对于V、V^2Bz、VB^2和ε，行星际电场的突变特征曲线更易识别；弱的扰动磁层背景状况和行星际磁场南向分量及电场晨昏分量的较大波动影响着磁暴的发展，使磁暴主相有多个发展阶段，从而增加磁暴的强度；对主相有多个发展阶段磁暴的研究有待进一步改善。     

GEO轨道相对论电子日积分通量预报统计建模

基于辐射带相对论电子哨声波局地加速理论,将地磁AE指数作为源电子通量和通量各向异性的指标,将地磁Dst指数作为损失机制的指标,利用滑动窗口线性滤波器方法,建立了一个地球静止轨道大于2MeV相对论电子预报模型.利用该模型开展了2000-2009年地球静止轨道相对论电子通量预报试验.研究发现,这10年总预报效率为0.818,2003年的预报效率（0.633）最低,2009年的预报效率（0.856）最高.模型预报效果与持续模型相比有很大提高,略低于利用太阳风参数作为输入的同类预报模型的预报效果.这说明即使在缺少太阳风参数的情况下,该模型利用地磁扰动参数也能取得较好的预报效果.当模型输入参数增加了太阳风速度时,即综合考虑了行星际扰动和磁层扰动对辐射带粒子加速过程的影响,模型逐年的预报效率进一步提升.其中,2005年的预报效率提升了9.5%,这10年的总预报效率增加到0.848,预报值与实测值之间的线性相关系数为0.918,均方根误差为0.422.      

行星际扰动与不同级别磁暴强度关系的研究

利用1997-2004年间ACE卫星太阳风观测的时均值数据和相应的Dst指数,针对Dstmin≤-50 nT的磁暴,分析了行星际参数(Bz,Ey,v,Pk,|B|,ε'=vxB2zsin4(θ/2))与Dst指数的相关关系.验证了Ey,Bz与Dst指数的良好相关性;按磁暴强度的不同,发现磁暴强度越大,行星际参数与磁暴强度(Dstmin)的相关性就越好.对于中等磁暴(-100 nT＜Dstmin≤-50 nT),行星际参数与磁暴强度的相关系数不高.如果把磁暴分为两个档次,即-150 nT＜Dstmin≤-50 nT的磁暴和Dstmin≤-150 nT的磁暴,计算结果表明,ε'与Dst指数的相关性是最好的.在诸多行星际参数中,就单一因素来说,Ey对磁暴强度影响最大,Bz对磁暴强度影响次之.      

基于深度学习递归神经网络的电离层总电子含量经验预报模型

利用行星际太阳风参数与太阳活动指数、地磁活动指数、电离层总电子含量格点化地图数据,首次基于一种能处理时间序列的深度学习递归神经网络（Recurrent Neural Network,RNN）,建立提前24h的单站电离层TEC预报模型.对北京站（40°N,115°E）的预测结果显示,RNN对扰动电离层的预测误差低于反向传播神经网络（Back Propagation Neural Network,BPNN）0.49～1.46TECU,将太阳风参数加入预报因子模型后对电离层正暴预测准确率的提升可达16.8%.RNN对2001和2015年31个强电离层暴预报的均方根误差比BPNN低0.2TECU,将太阳风参数加入RNN模型可使31个事件的平均预报误差降低0.36～0.47TECU.研究结果表明深度递归神经网络比BPNN更适用于电离层TEC的短期预报,且在预报因子中加入太阳风数据对电离层正暴的预报效果有明显改善.      

冕洞特征参数与地磁暴强度及发生时间统计

地磁暴是空间天气预报的重要对象.在太阳活动周下降年和低年,冕洞发出的高速流经过三天左右行星际传输到达地球并引发的地磁暴占主导地位.目前地磁暴的预报通常依赖于1AU处卫星就位监测的太阳风参数,预报提前量只有1h左右.为了增加地磁暴预报提前量,需要从高速流和地磁暴的源头即太阳出发,建立冕洞特征参数与地磁暴的定量关系.分析了2010年5月到2016年12月的152个冕洞-地磁暴事件,利用SDO/AIA太阳极紫外图像提取了两类冕洞特征参数,分析了其与地磁暴期间ap,Dst和AE三种地磁指数的统计关系,给出冕洞特征参数与地磁暴强度以及发生时间的统计特征,为基于冕洞成像观测提前1～3天预报地磁暴提供了依据.      

利用人工神经网络预报大磁暴

本文采用阈值预报的策略和人工神经网络BP模型,以13个太阳风参量和地磁AE,Dst指数作为输入,以0或1作为输出,提前4h预报大磁暴主相发生的时刻．结果表明,采用神经网络方法的阈值预报可以对灾害性磁暴的发生提前数小时做出比较准确的预报．     

利用神经网络技术提前一小时预报Dst指数

利用神经网络技术并考虑地磁活动的周期性,提出了一种提前一小时预报Dst指数的方法.网络的输入包括时间、季节、当前时刻及其一阶增量、二阶增量、前27天Dst指数的平均值.以下一时刻Dst指数作为输出对网络进行训练,训练好的网络可以提前一小时预报Dst指数.分别用1985年、1986年、1990年和1991年Dst指数数据进行检验.结果表明,预报结果与观测数据符合较好,Dst指数预报误差的均方根分别为4.00 nT,3.72 nT,5.35 nT,6.82 nT.误差分析表明,Dst指数的预报结果太阳活动低年比高年好.     

地球同步轨道高能电子增强事件预报方法

分析了地球同步轨道高能电子通量增强事件的发生规律及其与太阳风和行星际磁场参数的关系，并在此基础上建立了基于人工神经网络的高能电子增强事件模式，经实测数据检验，预报模式可以对未来1天的高能电子通量进行预报，误差为8．2％，达到了较高水平．     

利用转移熵研究引起磁暴扰动的太阳风参数重要性排序

磁暴是重要空间天气灾害性事件,能够影响卫星的安全在轨运行和地面电网系统等。目前,对于太阳风–磁层相互作用的研究多集中在分析相关系数的线性关系,而基于信息论的转移熵可以提供强大的无模型有向统计量,可用来分析传统相关性分析和模型假设检测不到的非线性关系。本文利用转移熵的方法,研究了磁暴期间的太阳风驱动参数。利用第23和24太阳活动周的小时精度数据进行长时间尺度分析,发现太阳风向地磁的信息传递呈双峰分布,表现出与太阳活动水平的一致性。利用2010-2018年93个地磁暴期间的分钟精度数据进行短时间尺度分析,结果表明：行星际电场(E)和行星际磁场南向分量(B z)对地磁指数Sym-H在时间延迟为60 min时信息传递较强,而太阳风速度v_{s w}、温度T _sw、数密度D_sw、磁场B和动压P_sw对Sym-H指数的信息传递较弱。上述研究结果能够为太阳风–磁层相互作用的建模提供参数选择及确定预测范围的依据。     

Forecasting of DST variations from polar cap indices using neural networks

The 15-min averaged polar cap (PC) index was used as an input parameter for the Dst variation forecasting. The PC index is known to describe well the principal features of the solar wind as well as the total energy input to the magnetosphere. This allowed us to design a neural network able to forecast the Dst variations from 1 to 4 h ahead. 1998 PC and Dst data sets were used for training and testing and 1997 data sets was used for validation proposes. From the 15 moderate and strong geomagnetic storms observed during 1997, nine were successfully forecasted. In three cases the observed minimum Dst value was less than the predicted one, and only in three cases the neural network was not able to reproduce the features of the geomagnetic storm.     

Evolution of Dst index,cosmic rays and global temperature during solar cycles 20–23

We have studied conditions in interplanetary space, which can have an influence on galactic cosmic ray (CR) and climate change. In this connection the solar wind and interplanetary magnetic field parameters and cosmic ray variations have been compared with geomagnetic activity represented by the equatorial Dst index from the beginning 1965 to the end of 2012. Dst index is commonly used as the solar wind–magnetosphere–ionosphere interaction characteristic. The important drivers in interplanetary medium which have effect on cosmic rays as CMEs (coronal mass ejections) and CIRs (corotating interaction regions) undergo very strong changes during their propagation to the Earth. Because of this CMEs, coronal holes and the solar spot numbers (SSN) do not adequately reflect peculiarities concerned with the solar wind arrival to 1 AU. Therefore, the geomagnetic indices have some inestimable advantage as continuous series other the irregular solar wind measurements. We have compared the yearly average variations of Dst index and the solar wind parameters with cosmic ray data from Moscow, Climax, and Haleakala neutron monitors during the solar cycles 20–23. The descending phases of these solar cycles (CSs) had the long-lasting solar wind high speed streams occurred frequently and were the primary contributors to the recurrent Dst variations. They also had effects on cosmic rays variations. We show that long-term Dst variations in these solar cycles were correlated with the cosmic ray count rate and can be used for study of CR variations. Global temperature variations in connection with evolution of Dst index and CR variations is discussed.     

Advances in space research nonlinear dependence of Dst and AE indices on the electric field of magnetic clouds

Using the Dst and AE geomagnetic index values and parameters of interplanetary magnetic field and solar wind we have examined the geoeffectiveness of transient ejections in the solar wind, namely, magnetic clouds and high-speed streams. It is found that for magnetic clouds the dependences of indices on the solar wind electric field are nonlinear of different kind. In contrast to magnetic clouds, the dependence of Dst and AE geomagnetic index values on the solar wind electric field agrees closely with the linear one for high-speed streams. We suggest approximating formulas to describe dependences obtained taking into account the relation of the electric field transpolar potential to the electric field and dynamic pressure of the solar wind. We suppose that the interplanetary magnetic field fluctuations also contribute to these dependences.     

The Earth’s magnetosphere response to interplanetary medium conditions on January 21–22, 2005 and on December 14–15, 2006

The Earth’s magnetosphere response to interplanetary medium conditions on January 21–22, 2005 and on December 14–15, 2006 has been studied. The analysis of solar wind parameters measured by ACE spacecraft, of geomagnetic indices variations, of geomagnetic field measured by GOES 11, 12 satellites, and of energetic particle fluxes measured by POES 15, 16, 17 satellites was performed together with magnetospheric modeling based in terms of A2000 paraboloid model. We found the similar dynamics of three particle populations (trapped, quasi-trapped, and precipitating) during storms of different intensities developed under different external conditions: the maximal values of particle fluxes and the latitudinal positions of the isotropic boundaries were approximately the same. The main sources caused RC build-up have been determined for both magnetic storms. Global magnetospheric convection controlled by IMF and substorm activity driven magnetic storm on December 14–15, 2006. Extreme solar wind pressure pulse was mainly responsible for RC particle injection and unusual January 21, 2005 magnetic storm development under northward IMF during the main phase.     

Storm-time atmospheric density modeling using neural networks and its application in orbit propagation

Upper atmospheric densities during geomagnetic storms are usually poorly estimated due to a lack of clear understanding of coupling mechanisms between the thermosphere and magnetosphere. Consequently, the orbit determination and propagation for low-Earth-orbit objects during geomagnetic storms have large uncertainties. Artificial neural networks are often used to identify nonlinear systems in the absence of rigorous theory. In the present study, an attempt has been made to model the storm-time atmospheric density using neural networks. Considering the debate over the representative of geomagnetic storm effect, i.e. the geomagnetic indices ap and Dst, three neural network models (NNM) are developed with ap, Dst and a combination of ap and Dst respectively. The density data used for training the NNMs are derived from the measurements of the satellites CHAMP and GRACE. The NNMs are evaluated by looking at: (a) the mean residuals and the standard deviations with respect to the density data that are not used in training process, and (b) the accuracy of reconstructing the orbits of selected objects during storms employing each model. This empirical modeling technique and the comparisons with the models NRLMSIS-00 and Jacchia-Bowman 2008 reveal (1) the capability of neural networks to model the relationship between solar and geomagnetic activities, and density variations; and (2) the merits and demerits of ap and Dst when it comes to characterizing density variations during storms.     

Interplanetary shocks and geomagnetic activity during solar maximum (2000) and solar minimum (1995–1996)

Plasma and magnetic field parameter variations through fast forward interplanetary shocks were correlated with the peak geomagnetic activity index Dst in a period from 0 to 3 days after the shock, during solar maximum (2000) and solar minimum (1995–1996). Solar wind speed (V) and total magnetic field (B_t) were the parameters with higher correlations with peak Dst index. The correlation coefficients were higher during solar minimum (r² = 56% for V and 39% for B_t) than during solar maximum (r² = 15% for V and 12% for B_t). A statistical distribution of geomagnetic activity levels following interplanetary shocks was obtained. It was observed that during solar maximum, 36% and 28% of interplanetary shocks were followed by intense (Dst  −100 nT) and moderate (−50  Dst < −100 nT) geomagnetic activity, whereas during solar minimum 13% and 33% of the shocks were followed by intense and moderate geomagnetic activity. It can be concluded that the upstream/downstream variations of V and B_t through the shocks were the parameters better correlated with geomagnetic activity level, and during solar maximum a higher relative number of interplanetary shocks can be followed by intense geomagnetic activity than during solar minimum. One can extrapolate, for forecasting goals, that during a whole solar cycle a shock has a probability of around 50% to be followed by intense/moderate geomagnetic activity.     

1997年1月6—11日日地连接事件：对磁暴—环电流—对流电场的讨论

给出了1997年1月6—11日日地连接事件的太阳风和行星际扰动及由此产生的地磁扰动特征．利用这些资料对磁暴-环电流-对流电场的分析表明,磁暴主相（或环电流）的开始主要是IMF南向分量形成的对流电场直接驱动的结果;对流电场在磁暴主相的形成中有极为重要的作用;但在主相发展的不同阶段作用不同