Forecasting space weather: Can new econometric methods improve accuracy?

Space weather forecasts are currently used in areas ranging from navigation and communication to electric power system operations. The relevant forecast horizons can range from as little as 24 h to several days. This paper analyzes the predictability of two major space weather measures using new time series methods, many of them derived from econometrics. The data sets are the A_p geomagnetic index and the solar radio flux at 10.7 cm. The methods tested include nonlinear regressions, neural networks, frequency domain algorithms, GARCH models (which utilize the residual variance), state transition models, and models that combine elements of several techniques. While combined models are complex, they can be programmed using modern statistical software. The data frequency is daily, and forecasting experiments are run over horizons ranging from 1 to 7 days. Two major conclusions stand out. First, the frequency domain method forecasts the A_p index more accurately than any time domain model, including both regressions and neural networks. This finding is very robust, and holds for all forecast horizons. Combining the frequency domain method with other techniques yields a further small improvement in accuracy. Second, the neural network forecasts the solar flux more accurately than any other method, although at short horizons (2 days or less) the regression and net yield similar results. The neural net does best when it includes measures of the long-term component in the data.     

Neural network development for the forecasting of upper atmosphere parameter distributions

This paper presents a neural network modeling approach to forecast electron concentration distributions in the 150–600 km altitude range above Arecibo, Puerto Rico. The neural network was trained using incoherent scatter radar data collected at the Arecibo Observatory during the past two decades, as well as the Kp geomagnetic index provided by the National Space Science Data Center. The data set covered nearly two solar cycles, allowing the neural network to model daily, seasonal, and solar cycle variations of upper atmospheric parameter distributions. Two types of neural network architectures, feedforward and Elman recurrent, are used in this study. Topics discussed include the network design, training strategy, data analysis, as well as preliminary testing results of the networks on electron concentration distributions.     

第22太阳周地磁活动峰值和时间的预计

本文讨论了从第13—22太阳周太阳和地磁周的特征.运用自激励门限自回归时间序列模型和最大熵谱原理自回归数学方法来模拟和预报地磁aa指数年均值峰值和时间.峰值时间是1993年秋天或1994年春天.地磁aa指数年均值峰值是26—29.第22地磁周是一个中等活动的周.     

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.     

基于机器学习相似度算法的Kp指数预报

基于机器学习中的相似度算法,建立了在历史太阳风数据中寻找与当前太阳风特征相近事例的推荐模型,用来预报地磁Kp指数.使用1998-2019年间随机选择的120个太阳风事例作为测试数据集,该模型能够推荐得到历史上与输入太阳风造成相似地磁影响的太阳风事例,最优事例的Kp指数与实际值的均方根误差为0.79,相关系数为0.93....     

Forecast of recurrent geomagnetic storms

Long-term forecast of space weather allows in achieving a longer lead time for taking the necessary precautions against disturbances. Hence, there is a need for long-term forecasting of space weather. We studied the possibility for a long-term forecast of recurrent geomagnetic storms. Geomagnetic storms recur with an approximate 27-day period during the declining phase of a solar cycle. These disturbances are caused by the passage of corotating interaction regions, which are formed by interactions between the background slow-speed solar wind and high-speed solar wind streams from a coronal hole. In this study, we report on the performance of 27-day-ahead forecasts of the recurrent geomagnetic disturbances using Kp index. The methods of the forecasts are on the basis of persistence, autoregressive model, and categorical forecast using occurrence probability. The forecasts show better performance during the declining phase of a solar cycle than other phases. The categorical forecast shows the probability of detection (POD) more than 0.5 during the declining phase. Transition of the performance occurs sharply among the declining phases and other phases.     

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．     

On the mid-latitude ionospheric storm association with intense geomagnetic storms

The bulk association between ionospheric storms and geomagnetic storms has been studied. Hemispheric features of seasonal variation of ionospheric storms in the mid-latitude were also investigated. 188 intense geomagnetic storms (Dst ≤ 100 nT) that occurred during solar cycles 22 and 23 were considered, of which 143 were observed to be identified with an ionospheric storm. Individual ionospheric storms were identified as maximum deviations of the F2 layer peak electron density from quiet time values. Only ionospheric storms that could clearly be associated with the peak of a geomagnetic storm were considered. Data from two mid-latitude ionosonde stations; one in the northern hemisphere (i.e. Moscow) and the other in the southern hemisphere (Grahamstown) were used to study ionospheric conditions at the time of the individual geomagnetic storms. Results show hemispheric and latitudinal differences in the intensity and nature of ionospheric storms association with different types of geomagnetic storms. These results are significant for our present understanding of the mechanisms which drive the changes in electron density during different types of ionospheric storms.     

不同速度运动汽车的磁异信号研究与探测

运动汽车随着速度的增加伴随着非稳态材料的产生，如汽油高温燃烧、高速摩擦引起自由电荷累积等。针对这些磁异信号难以采用适用于铁磁性材料的磁偶极子模型描述的问题，提出了结合磁偶极子模型和运动电荷等效模型的方法，理论计算不同速度运动汽车的磁异信号并分析其时域、频域特征，获得磁异信号与速度的依赖关系。采用隧道磁阻传感器（TMR）结合滤波、放大、模数转换技术构建弱磁信号探测实验装置，探测不同速度运动汽车的时域磁异信号，并采用傅里叶变换获得其频域信息，与理论模型相吻合。随着速度的增加，频域信号向高频方向偏移，对于从低频地磁背景场中提取目标弱磁信号极其重要。      

利用行星际监测数据进行地磁暴预报

利用全连接神经网络方法应用于地磁Dst指数的预报中.对ACE卫星探测的太阳风和行星际磁场及其变化对未来几小时的Dst指数的影响进行了统计分析,发现在这些行星际实测参数中,对Dst指数作用较为明显的是太阳风速度、太阳风质子密度和行星际磁场南向分量,同时,当前Dst指数实测值对今后几小时的Dst指数已有很强的制约作用.在统计分析的基础上,建立了全连接神经网络预报模型.由于采用了全连接神经网络结构,模式能够反映出太阳风、行星际磁场等参数与地磁Dst指数参数的复杂联系,可以自动建立输入参量的最佳组合方式,提高了预报精度.通过利用大量实测数据对神经网络模式进行训练,最终建立了利用优选的ACE卫星行星际监测数据提前2 h对Dst指数进行预报.通过检测,预报的误差为14.3%.      

一种高动态环境下卫星扩频信号的快速捕获方法研究

现代卫星导航及测控应用对接收机在高动态环境下实现测量通信提出了迫切需求。为了解决大多普勒频偏扩频信号的快速捕获问题,提出了一种在频域并行搜索码相位及多普勒频偏的双频域快速捕获方法。采用双块补零算法将长的相关积分操作分割为多个短的相关积分操作,然后采用快速傅里叶变换进行圆周相关,大大节约了处理时间,利用频域圆周移位与时域载波剥离等价的原理,大幅提高了频率搜索效率。与时域相关算法和单频域计算方法相比,在捕获灵敏度不变的条件下,该方法将计算量减少90%,显著提高了运算速度,适合高动态环境下扩频信号的快速捕获。该方案应用于星载接收机平台FPGA实现,测试结果表明该方案可以在0.1s内完成±500kHz频偏下扩频信号捕获。     

Hemispheric Distribution of Lower-band Chorus Waves Observed by Van Allen Probes

Whistler mode chorus waves are important electromagnetic emissions due to their dual roles in acceleration and loss processes of Earth’s radiation belt electrons. A detailed global survey of lower-band chorus is performed using EMFISIS data from Van Allen Probes in near-equatorial orbits. In addition to the confirmation of the positive correlation of chorus wave intensities to geomagnetic activity and dayside-nightside distribution asymmetry of wave amplitude and occurrence probability, the analysis results find that in statistics lower-band chorus emissions exhibit higher wave occurrence rates and larger normalized peak wave frequencies in the magnetically northern hemisphere but somehow stronger peak wave intensities in the magnetically southern hemisphere. While overall the differences between the two magnetically hemispheric distributions tend to be not significant, it is important to establish the magnetically hemispheric distribution profiles of lowerband chorus with respect to L-shell, magnetic local time, and geomagnetic latitude for improved understanding of chorus-induced dynamics of radiation belt electrons.      

冕洞相关地磁Ap指数中短期预报方法研究

地磁Ap指数是描述全球地磁活动水平的重要指数, 过去许多参考大气模式中都用Ap指数来表述地磁活动状态, 大气模式的运行需要输入地磁Ap指数, 因此, 地磁Ap指数的预报一直是空间环境预报中一个非常重要的内容. 针对太阳活动低年冕洞引起的地磁扰动具有明显27天重现的特性, 利用修正的自回归方法, 对地磁Ap指数进行了提前27天的预报; 采用从SOHO/EIT观测资料发展出来的描述冕洞特性的Pch因子, 进行了提前三天的地磁Ap指数预报. 结果显示, 将统计方法与物理分析相结合, 进行地磁Ap指数的中短期数值预报, 可以得到较好的预报效果.      

基于多伪码周期的直扩快速捕获新方法

一般高动态扩频接收系统中用于快速捕获的FFT运算主要是在频域或者码域中的一个域进行,在另一个域中步进搜索,而步进造成捕获时间成倍增加。通过在一个数据周期内嵌入一定数量的伪码周期,提出一种能同时在频域和码域中应用FFT进行并行捕获的新方法。仿真表明该方法捕获时间短、精度高。     

一种基于时间序列的自适应网络异常检测算法

传统的网络管理工具通常是根据预先设定的阈值进行网络流量异常检测,这种方法虽然简单,但不能根据网络状况进行自适应的动态调整.分析了基于时间序列的Holt-Winters异常检测方法,结合建立的历史流量的正常模型,改进了Holt-Winters模型的基值以及平滑因子参数的获取过程,加快了算法的启动时间,缩短了算法对网络环境的自适应时间.改进的Holt-Winters算法相较于原来的Holt-Winters算法以及阈值检测方法检测的正确率更高、误报率更低.     

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.     

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.      

宇宙线变化特征在强地磁暴预报中的应用

利用宇宙线中子探测数据定性分析了地面宇宙线多台站之间的相互联系以及大磁暴与宇宙线之间的响应关系. 以Irkutsk和Oulu宇宙线台站为例, 运用小波去噪技术提高数据的稳定性. 结果表明, 相同世界时条件下, 两站宇宙线通量相关性在事件发生时较高; 而相同地方时条件下, 相关性则在平静期较高. 进一步采用相同地方时条件对不同宇宙线台站的通量在平静期和扰动期的相对变化进行分析, 选取2004年7月强地磁暴典型事例进行直观分析, 发现大地磁暴前Irkutsk和Oulu台站的宇宙线相对通量发生明显差异, 可以尝试作为强地磁暴宇宙线先兆特征. 通过对2001年3月至2005年5月的强磁暴和中强磁暴进行统计, 得到与强地磁暴相关的适当宇宙线相对差异阈值. 将得到的阈值对2005年9月至2011年12月所有强磁暴及中强磁暴进行验证, 总成功率达到87.5%, 误报率为35.7%, 结果较好.      

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。     

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.