电离层暴时经验模型STORM在中国区域的适应性研究

利用中国区域内9个垂测站1976---1987年一个太阳活动周期的电离层暴时f0F2数据, 统计分析了电离层暴事件的等级, 以及不同等级的电离层暴随季节和地磁纬度的分布特征. 研究发现, 中小型电离层暴在春秋季发生的概率较大, 不同季节的发生次数与地磁纬度具有明显的关系. 利用STORM模型对电离层暴时f0F2和大型及特大型电离层暴时f0F2的预测值与月中值进行了比较. 结果表明, 除了冬季误差增大外, 发生电离层暴时STORM模型能够有效地改善月中值模型. 增加中国的暴时数据, 并提高对冬季的暴时参数f0F2的预测是改善STORM模型的重要因素. 建立合适的暴时指数来预测f0F2是未来研究的重点.      

地磁扰动期间日本Kokubunji站电离层的扰动特征分析

利用日本Kokubunji站(139.5°E,35.5°N)1959年1月到2004年12月共46年的F2层临界频率foF2参数,统计分析了Kokubunji站电离层F2层峰值电子浓度NmF2随地磁活动、太阳活动、季节和地方时变化的形态特征.结果表明,总体来看,磁暴期间Kokubunji站电离层响应以正暴为主,其中在太阳高年夏季为负暴,冬季为正暴,春秋季以负暴为主但幅度较小;在太阳低年夏季以正暴为主,冬季为正暴,春秋季以正暴为主.NmF2扰动与ap指数在夏季太阳高年负相关,在冬季无论太阳高年低年均为正相关,春秋季中4月和9月在太阳高年类似夏季,3月和10月在太阳低年类似冬季.电离层最大负相扰动对最大地磁活动的延迟时间约为12～15 h;正相扰动的延迟时间则分别为3 h和10 h.地磁活跃期间地方时黄昏后到午夜前倾向于正相扰动,清晨倾向于负相扰动.      

集合卡尔曼滤波在电离层短期预报中的应用

提出了一种利用集合卡尔曼滤波对电离层f0F2短期预报结果进行优化的方法. 利用训练好的神经网络对f0F2进行提前1~24 h的预报, 考虑前一天预报误差的反馈信息, 动态跟踪 f0F2的变化趋势, 引入集合卡尔曼滤波对神经网络的预报结果实行进一步修正和优化. 实验结果表明, 此方法的预报效果优于单纯的神经网络模型和IRI模型. 此方法还可以应用于其他电离层参量的短期预报.      

相似预报法在电离层TEC短期预报中的应用

引入相似离度衡量样本间的相似程度, 并利用相似预报法对厦门一个GPS台站2004年电离层TEC观测数据进行了24,h预报试验. 结果表明, 预报相对误差与地磁活动水平密切相关, 地磁扰动条件下相对误差明显高于地磁平静时刻; 预报相对平均误差为18.022%, 地磁扰动时为44.896%, 地磁平静条件下为11.676%; 预报相对误差在10%, 20%, 30%, 40%以内的累积比例分别为38.209%, 65.075%, 84.984%,90.448%. 如果使用中纬地区或地磁平静期间的电离层TEC数据, 预报效果会更好.      

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.      

F2层对地磁扰动的响应

利用37个电离层垂直探测站1974-1986年的数据,采用f0F2与地磁ap指数相关分析的方法,首次得到一个太阳活动周期各年东亚-澳大利亚扇区,欧洲-非洲扇区和美洲-东太平洋扇区F2层对地磁扰动响应随地磁纬度的分布.结果指出,地磁高纬和中纬地区为负响应,低纬和赤道地区为正响应,大约在±30°附近换向.最大正响应在磁赤道附近,最大负响应在地磁纬度±50°附近,最大负响应的幅度大于最大正响应的幅度.存在明显的经度差别和南北半球不对称性.     

地磁活动对电离层f0F2月中值的影响与国际参考电离层误差

利用我国9个中低纬度的电离层观测站在1977-1986年间观测的f0F2月中值,按每月的平均地磁活动指数Ap分为地磁活动高（Ap≥5）和低（Ap＜15）两种情况,研究了地磁活动对f0F2月中值平均低纬电离层驼峰区演变的影响,并考察了国际参考电离层（IRI）的误差．     

利用神经网络预报电离层f0F2

由中国武汉电离层台站和澳大利亚Hobart台站的电离层F2层临界频率(f0F2)的资料,利用三层前向反馈神经网络(BP网络),提出一种提前24h预测f0F2的方法,该方法以前5天观测的f0F2数据拟合的5个系数以及太阳活动参数作为输入,以当天24 h的f0F2作为输出对网络进行训练,训练好的网络可以实现对f0F2提前24 h的预报.预测结果显示,利用神经网络预测的f0F2与实际观测结果变化趋势较一致,并且比IRI的计算结果更加准确.误差分析表明,在南半球Hobart(-42.9°,147.3°)台站比中国武汉站(30.4°,114.3°)的结果要好,在低年比高年要好,在冬夏季节比春秋季节稍好.本文说明利用神经网络对电离层参量进行预报是一种切实可行的方法.     

基于因果卷积与LSTM网络的电离层总电子含量预报

电离层总电子含量（TEC）不仅是分析电离层形态的关键参数之一,同时为导航及定位等空间应用系统消除电离层附加时延提供重要支撑。由于电离层TEC的时空变化特征,本文融合因果卷积和长短时记忆网络,以太阳活动指数F_10.7、地磁活动指数Dst和电离层TEC历史数据作为特征输入,构建深度学习模型,实现提前24 h预报电离层TEC。进一步利用2005－2013年连续9年的CODE TEC数据,全面评估了模型在北京站（40°N,115°E）、武汉站（30.53°N,114.36°E）和海口站（20.02°N,110.38°E）的预报性能。结果显示不同太阳活动条件下三个站的TEC值与真实测量值的相关系数都大于0.87,均方根误差大都集中在0～1 TECU以内,且模型预报精度与纬度、太阳、地磁活动程度、季节变化相关。与仅由长短时记忆网络构成的预报模型相比,本实验模型均方根误差降低了15%,为电离层TEC预报模型的实际应用提供了参考。      

利用F_(10.7)短期预报电离层F_0F_2

通过对电离层历史数据和太阳射电流量F10.7的回归分析,提出了一种单站电离层f0F2的短期预报方法,以F10.7的流动平均值fc为输入,以未米3天的f0F2为输出,分别利用中国地区8个台站的数据进行检验,分析不同太阳活动水平、季节以及地方时预报误差的分布特征.结果表明,该方法能有效地预测未来1～3天的f0F2.该方法还可应用于其他电离层参量的短期预报.     

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.     

CME引起的地磁暴穿越时间

日冕物质抛射(CME)从发生至引起地磁暴最大值的时间间隔称为穿越时间.本文选取1997-2015年89个CME-Dst事件,分析CME速度、能量、耀斑类型等对穿越时间的影响;采用非线性拟合以及支持向量机(SVM)非线性回归技术,建立基于1997-2009年62个CME-Dst事件的CF模型和SVM模型,并利用其余27个CME-Dst事件对模型预报效果分别进行检验.结果表明,CF模型和SVM模型的预报准确率均达到85.2%,其中CF模型的平均绝对值误差为13.77 h,而SVM模型为13.88 h.与ECA模型结果(准确率为77.8%,平均绝对值误差为14.55 h)进行对比发现,CF模型和SVM模型的准确率更高而误差更小.CF模型和SVM模型能够提前1~5天较好地预报地磁暴爆发时间.     

Mid-latitude thermospheric wind changes during the St. Patrick’s Day storm of 2015 observed by two Fabry-Perot interferometers in China

In this work, we utilize thermospheric wind observations by the Fabry-Perot interferometers (FPI) from the Kelan (KL) station (38.7°N, 111.6°E, Magnetic Latitude: 28.9°N) and the Xinglong (XL) station (40.2°N, 117.4°E, Magnetic Latitude: 30.5°N) in central China during the St. Patrick’s Day storm (from Mar. 17 to Mar. 19) of 2015 to analyze thermospheric wind disturbances and compare observations with the Horizontal Wind Model 2007 (HWM07). The results reveal that the wind measurements at KL show very similar trends to those at XL. Large enhancements are seen in both the westward and equatorward winds after the severe geomagnetic storm occurred. The westward wind speed increased to a peak value of 75?m/s and the equatorward wind enhanced to a peak value of over 100?m/s. There also exist obvious poleward disturbances in the meridional winds during Mar. 17 to Mar. 19. According to the comparison with HWM07, there exist evident wind speed and temporal differences between FPI-winds and the model outputs in this severe geomagnetic storm. The discrepancies between the observations and HWM07 imply that the empirical model should be used carefully in wind disturbance forecast during large geomagnetic storms and more investigations between measurements and numerical models are necessary in future studies.     

基于经验模态分解的地球同步轨道高能电子通量预报

在磁暴恢复相期间,大量相对论(高能)电子从磁层的外辐射带渗透到地球同步轨道区.其中> 2 MeV的高能电子能够穿透卫星表面并聚积在材料内部,导致卫星无法正常运行或完全损坏.磁暴期间的高能电子通量变化的非平稳与非线性特征十分明显.通过实验发现,经验模态分解法能够极大地降低高能电子通量非平稳性问题造成的预报影响.以2008-2009年的数据作为训练集,2010-2013年数据作为测试集.结果表明:2010-2013年的预报率约为0.84;在太阳活动较为复杂的2013年,预报率达到0.81.引入经验模态分解后预报效率得到显著提高.     

基于经验加速度的低轨卫星轨道预报新方法

研究将定轨过程中的经验加速度应用于地球低轨卫星轨道预报的新方法. 利用GPS伪距观测数据和简化动力学最小二乘批处理方法对地球低轨卫星定 轨, 其中卫星位置、速度及大气阻力系数和辐射光压系数可以直接用于轨道预报. 作为简化动力学最重要特征的经验加速度呈现准周期、余弦曲线特点, 可通过 傅里叶级数拟合建模. 确定性动力学模型与补偿大气阻力模型误差的切向经验 加速度级数拟合模型组成增强型动力学模型用于提高轨道预报精度. 应用 GRACE-A星载GPS伪距观测数据和IGS超快星历定轨并进行轨道预报, 结果表明 轨道预报初值位置精度达到0.2m, 速度精度达到1×10-4m·s-1, 预报3天位置精度优于60m, 比只利用确定性动力学模型进行预报精度平 均提高2.3倍. 先定轨后预报的模式可用在星上自主精确导航系统中.      

一种基于温度参数的热层密度修正方法

热层大气的阻力效应是影响低轨航天器大量空间操作的重要因素, 尤其是经验密度模式, 其固有的至少15%的内符合误差已严重制约航天器轨道计算精度的提高. 针对广泛应用的经验密度模式, 选择物理背景简明、关联参数较少的JACCHIA71模式, 以地磁平静条件下的全球散逸层顶温度最小值T_c及125 km高度拐点温度T_x为对象, 建立密度相对于上述温度参数的条件方程, 推导密度相对于温度参数的解析偏导数, 并给出其最小二乘解. 同时, 利用CHAMP卫星数据对模式进行修正, 模式平均误差从40%降低至3%左右. 通过TG01飞行器的轨道预报比较, 修正前后轨道预报位置精度从2 km提升至1 km左右. 经过CHAMP卫星和TG01飞行器的实测数据检验, 验证了修正算法的正确性和有效性.      

Use of two-line element data for thermosphere neutral density model calibration

Traditional empirical thermospheric density models are widely used in orbit determination and prediction of low-Earth satellites. Unfortunately, these models often exhibit large density errors of up to around 30% RMS. Density errors translate into orbit errors, adversely affecting applications such as re-entry operations, manoeuvre planning, collision avoidance and precise orbit determination for geodetic missions. The extensive database of two-line element (TLE) orbit data contains a wealth of information on satellite drag, at a sufficiently high spatial and temporal resolution to allow a calibration of existing neutral density models with a latency of one to two days. In our calibration software, new TLE data for selected objects is converted to satellite drag data on a daily basis. The resulting drag data is then used in a daily adjustment of density model calibration parameters, which modify the output of an existing empirical density model with the aim of increasing its accuracy. Two different calibration schemes have been tested using TLE data for about 50 objects during the year 2000. The schemes involve either height-dependent scale factors to the density or corrections to CIRA-72 model temperatures, which affect the density output based on a physical model. Both schemes have been applied with different spherical harmonic expansions of the parameters in latitude and local solar time. Five TLE objects, varying in perigee altitude between 280 and 530 km, were deliberately not used during calibration, in order to provide independent validation. Even with a single daily parameter, the RMS density model error along their tracks can already be reduced from the 30% to the 15% level. Adding additional parameters results in RMS errors lower than 12%.     

An empirical relation to correct storm-time thermospheric mass density modeled by NRLMSISE-00 with CHAMP satellite air drag data

With the help of STAR (Spatial Triaxial Accelerometer for Research) accelerometer measurements on board CHAMP (Challenging Minisatellite Payload), the global distributions of total mass density changes at about 400 km height during major magnetic storms are studied, aiming to improve the capability of current thermospheric model for predicting the storm-time mass density distribution. The density calculated by the NRLMSISE-00 model without using the geomagnetic active index as input is taken as a reference on top of which the storm-time changes are added. In total 19 storm events during 2001–2004 are used to perform a comprehensive statistical analysis. A relative calibration of drag coefficient along with accelerometer calibration parameters is made by fitting the CHAMP observed initial mass densities in with the NRLMSISE-00 model on quiet days before each storm. The dependences of the storm-time changes in mass density on both the total global Joule heating power, ∑_Qj$\sum Q_{\text{j}}$ and the high-resolution ring current index, Sym-H, are investigated. The lag times of mass density changes with respect to the Joule heating and Sym-H variation are obtained as a function of latitude and sunlight. By using a multiple linear regression analysis with proper time shift, an empirical relation connecting storm-time changes in mass density for 400 km height with the two parameters, ∑_Qj$\sum Q_{\text{j}}$ and Sym-H, has been worked out for different latitude and sunlight conditions (day-side or night-side). Adding a correction calculated from the empirical relation to the NRLMSISE-00 model reference leads to a better prediction of storm-time thermospheric mass density distribution.