基于SDO/HMI全日面实时磁图的活动区自动识别

太阳活动区是太阳活动的主要发生源区,活动区的形态、结构、特征是预报太阳爆发的主要依据.因此,活动区的识别是实现太阳爆发预报的前提.SDO/HMI能够提供连续、高时空精度的全日面光球观测图像.参照文献[1]SOHO/MDI综合磁图中活动区的自动识别方法,利用实时可得的HMI全日面磁图,通过阈值法、数学形态法和区域增长法相...     

基于活动区面积统计的F10.7指数预报方法

太阳10.7cm射电流量（F_10.7）是反映太阳整体活动的重要指标,其主要源头是日面活动区.F_10.7指数与日面活动区具有显著的相关性,且不同面积的活动区与F_10.7并不遵循相同的线性关系.为进一步提高F_10.7预报的准确性,利用日面活动区面积与F_10.7的相关性,依据面积大小分类,提出F_10.7的预报公式并进行验证.采用2008-2018年SWPC （Space Weather Prediction Center）公布的活动区面积数据和CSWFC （Canadan Space Weather Forecast Center）公布的F_10.7实测数据计算预报公式系数,利用高年（2003年）和低年（1997年）的F_10.7预报验证其结果.研究结果表明,预报结果与实测值的相关系数分别为0.9318和0.9295,二者皆优于SWPC同时期的预报结果（相关系数分别为0.9186和0.8771）.本研究首次基于活动区的变化预测了F_10.7,提高了F_10.7预测的准确性.      

太阳10.7 cm射电流量中期预报模型研究(Ⅱ)

太阳活动指数中期预报一直是空间环境业务预报的难点之一.本文在自回归方法模型的基础上,利用太阳活动区面积、位置等参数与10.7cm辐射流量之间的定量关系,根据活动区面积衰减规律,建立了一个基于活动区参数及演化规律的改进型太阳活动指数中期预报模型.通过对预报测试实例分析发现,在日面出现较大活动区导致F_(10.7)迅速增长并超过历史数据峰值的情况,在日面活动区消亡导致指数突然出现平静期的情况,新模型的预报准确性相比自回归模型有很大提高,预报的平均相对误差下降约5%～9%.由此可见,新模型在某些特定条件下提高了原有模型的精度.该研究为提高业务型太阳10.7cm射电流量中期预报模型的预报精度奠定了基础.     

基于长短期记忆神经网络的太阳耀斑短期预报

提出了一个基于长短期记忆神经网络的耀斑预报模型,利用过去24 h太阳活动区的磁场变化时序构建样本,通过长短期记忆神经网络对磁场特征时序演化进行分析,预报未来48 h内是否发生≥M级别耀斑事件。使用的数据集为2010年5月到2017年5月所有活动区样本,选取了SDO/HMI SHARP的10个磁场特征参量。在建模过程中通过XGBoost方法选取权重、增益率和覆盖率均较高的6个特征参量作为输入参数。通过测试对比,模型的虚报率和准确率与传统机器学习模型相近,报准率和临界成功指数分别为0.7483和0.7402,优于传统机器学习模型。模型总体效果优于传统机器学习模型。      

基于PCA方法对太阳活动区主要参量的分析

太阳耀斑与太阳质子事件的发生通常与太阳活动区存在非常密切的关系, 对这种关系的深入分析有助于太阳耀斑和太阳质子事件预报模型的建立. 本文利用主成分分析(Principal Component Analysis, PCA)方法对1997-2010年太阳质子事件所在活动区的主要参量进行分析, 选取的参量包括黑子磁分类、 McIntosh分类、太阳黑子群面积、10.7 cm射电流量、耀斑指数、质子耀斑位置和软X射线耀斑强度. 结果得到81个太阳活动主成分得分值排序(得分值代表每个事件的强弱), 与太阳质子事件峰值流量、太阳黑子年均值以及10.7 cm射电流量年均值的对比显示相似度非常高, 表明主成分得分值一定程度上可以反映太阳活动的强弱规律.      

利用相似周方法预测第24活动周开始时间

利用相似周方法对第24活动周的开始时间与第23活动周下降相后期的太阳黑子数进行了预报.根据第23周已经出现的特征参量和下降相的形态特征,选取9,10,11,15,17和20等六个太阳活动周作为第23周下降相的相似周,对第24周开始时间进行预报.预报结果显示,第24活动周的开始时间为2007年5±1月,黑子数平滑月均极小值为7.1±2.6,第23太阳活动周长度为11.1年.与其他研究者的预报结果相比较,本文给出的结果与文献[11]和[12]及MSFC的结果比较一致.通过对相似周方法在下降相预测太阳活动周结束时间的研究讨论,及对第23周上升阶段的太阳黑子数和F10.7平滑月均值预报结果的评估,可以看出,相似周预报方法在太阳活动周长期预报中是很有应用价值的.      

1986年2月太阳的高活动I活动区4711的演化和特征

本文使用太阳黑子、磁场、Hα色球、10.7cm射电及软X射线流量等观测资料,对太阳活动谷期的高活动区4711(SESC编号)从光球、色球和日冕三个方面做了综述.指出该活动区演化过程的特征是:(1)黑子群在主要发展阶段呈一个紧密的结构复杂的强磁区;(2)两次大的太阳爆发均发生在黑子群面积衰减阶段的初期;(3)黑子群的转动可能是活动区日冕加热和耀斑活动的主要供能机制;(4)色球暗条的频繁活动是爆发的先兆;(5) 10.7cm射电辐射和软X射线辐射的逐日流量有彼此不重合的双峰.      

基于深度学习的太阳黑子Wilson山磁类型识别方法

太阳黑子是太阳光球层中带有较强磁场的区域,通常是太阳爆发活动的源区。Wilson山磁分类是当前最为主流的太阳黑子分类方法之一,对研究太阳爆发有重要意义。利用2010－2017年间SDO/HMI成像仪观测到的720s_SHARP磁图和白光图数据,研究使用深度学习对太阳黑子群Wilson山磁分类的方法。实验结果表明,Xception网络在识别太阳黑子Wilson山磁类型上能取得最优的效果,其中对α类型黑子的F₁得分为96.50%,β类为93.20%,其他类型的黑子为84.65%。      

LSTM神经网络在太阳F10.7射电流量中期预报中的应用

F_10.7指数作为大气密度经验模型的重要输入参量,其预报精度直接影响航天器轨道预报精度.研究发现,太阳活动表现出长时间尺度上平均11年和中短时间尺度平均27天的周期性变化特征.依据这一观测事实,基于长短期记忆单元（Long Short-term Memory,LSTM）递归神经网络方法进行F_10.7指数未来27天的中期预报.利用一个连续长时段F_10.7数据作为训练数据,构建LSTM神经网络训练和预测模型,分别预测太阳活动高低年未来27天的F_10.7指数.结果表明,太阳活动高年的第27天F_10.7指数预报平均相对误差最优可达10%以内,低年最优可达2%以内.      

用BP神经网络预报太阳活动第23周的黑子数

本文设计、训练和利用BP神经网络,对1750年以来的各太阳活动周上升段和下降段太阳黑子数的变化数据进行了分类和模式识别,得到各太阳活动周上升周期及其上升期间太阳黑子数平滑月均值相当好的模拟结果;在此基础上获得较好的太阳活动第22周上升周期及太阳黑子数的最大平滑月均值预报结果;还作出太阳活动第23周的上升周期及太阳黑子数的最大平滑月均值的预报结果.      

The automated prediction of solar flares from SDO images using deep learning

In the last few years, there has been growing interest in near-real-time solar data processing, especially for space weather applications. This is due to space weather impacts on both space-borne and ground-based systems, and industries, which subsequently impacts our lives. In the current study, the deep learning approach is used to establish an automated hybrid computer system for a short-term forecast; it is achieved by using the complexity level of the sunspot group on SDO/HMI Intensitygram images. Furthermore, this suggested system can generate the forecast for solar flare occurrences within the following 24 h. The input data for the proposed system are SDO/HMI full-disk Intensitygram images and SDO/HMI full-disk magnetogram images. System outputs are the “Flare or Non-Flare” of daily flare occurrences (C, M, and X classes). This system integrates an image processing system to automatically detect sunspot groups on SDO/HMI Intensitygram images using active-region data extracted from SDO/HMI magnetogram images (presented by Colak and Qahwaji, 2008) and deep learning to generate these forecasts. Our deep learning-based system is designed to analyze sunspot groups on the solar disk to predict whether this sunspot group is capable of releasing a significant flare or not. Our system introduced in this work is called ASAP_Deep. The deep learning model used in our system is based on the integration of the Convolutional Neural Network (CNN) and Softmax classifier to extract special features from the sunspot group images detected from SDO/HMI (Intensitygram and magnetogram) images. Furthermore, a CNN training scheme based on the integration of a back-propagation algorithm and a mini-batch AdaGrad optimization method is suggested for weight updates and to modify learning rates, respectively. The images of the sunspot regions are cropped automatically by the imaging system and processed using deep learning rules to provide near real-time predictions. The major results of this study are as follows. Firstly, the ASAP_Deep system builds on the ASAP system introduced in Colak and Qahwaji (2009) but improves the system with an updated deep learning-based prediction capability. Secondly, we successfully apply CNN to the sunspot group image without any pre-processing or feature extraction. Thirdly, our system results are considerably better, especially for the false alarm ratio (FAR); this reduces the losses resulting from the protection measures applied by companies. Also, the proposed system achieves a relatively high scores for True Skill Statistics (TSS) and Heidke Skill Score (HSS).     

Flux emergence,flux imbalance,magnetic free energy and solar flares

Emergence of complex magnetic flux in the solar active regions lead to several observational effects such as a change in sunspot area and flux embalance in photospheric magnetograms. The flux emergence also results in twisted magnetic field lines that add to free energy content. The magnetic field configuration of these active regions relax to near potential-field configuration after energy release through solar flares and coronal mass ejections. In this paper, we study the relation of flare productivity of active regions with their evolution of magnetic flux emergence, flux imbalance and free energy content. We use the sunspot area and number for flux emergence study as they contain most of the concentrated magnetic flux in the active region. The magnetic flux imbalance and the free energy are estimated using the HMI/SDO magnetograms and Virial theorem method. We find that the active regions that undergo large changes in sunspot area are most flare productive. The active regions become flary when the free energy content exceeds 50% of the total energy. Although, the flary active regions show magnetic flux imbalance, it is hard to predict flare activity based on this parameter alone.     

Application of support vector machine combined with K-nearest neighbors in solar flare and solar proton events forecasting

The support vector machine (SVM) combined with K-nearest neighbors (KNN), called the SVM-KNN method, is new classing algorithm that take the advantages of the SVM and KNN. This method is applied to the forecasting models for solar flares and proton events. For the solar flare forecasting model, the sunspot area, the sunspot magnetic class, and the McIntosh class of sunspot group and 10 cm solar radio flux are chosen as inputs; for the solar proton event forecasting model, the inputs include the longitude of active regions, the flux of soft X-ray, and those for the solar flare forecasting model. Detailed tests are implemented for both of the proposed forecasting models, in which the SVM-KNN and the SVM methods are compared. The testing results demonstrate that the SVM-KNN method provide a higher forecasting accuracy in contrast to the SVM. It also gives an increased rate of ‘Low’ prediction at the same time. The ‘Low’ prediction means occurrence of solar flares or proton events with predictions of non-occurrence. This method show promise for forecasting models of solar flare and proton events.     

Solar flare forecasting model supported with artificial neural network techniques

Nowadays operational models for solar activity forecasting are still based on the statistical relationship between solar activity and solar magnetic field evolution. In order to set up this relationship, many parameters have been proposed to be the measures. Conventional measures are based on the sunspot group classification which provides limited information from sunspots. For this reason, new measures based on solar magnetic field observations are proposed and a solar flare forecasting model supported with an artificial neural network is introduced. This model is equivalent to a person with a long period of solar flare forecasting experience.     

太阳活动对电离层TEC变化影响分析ormalsize

为研究太阳活动对电离层TEC变化的影响,从整体到局部分析了2000—2016年的太阳黑子数、太阳射电流量F_10.7指数日均值与电离层TEC的关系,并重点分析了2017年9月6日太阳爆发X9.3级特大耀斑前后15天太阳活动与电离层TEC变化的相关性.结果表明:由2000—2016年的数据整体看来,太阳黑子数、太阳F_10.7指数、TEC两两之间具有很强的整体相关性,但局部相关性强弱不均;此次耀斑爆发前后太阳黑子数、太阳F_10.7指数和TEC具有很强的正相关特性,太阳活动对TEC的影响时延约为2天;太阳活动对全球电离层TEC的影响不同步,从高纬至低纬约有1天的延迟,且对低纬度的影响远大于中高纬度.太阳活动是影响电离层TEC变化的主要原因,但局部也可能存在其他重要影响因素.