Preliminary Study on Paradigm Shift in Space Weather Research Driven by Big Data and Artificial Intelligence

doi:10.11871/jfdc.issn.2096-742X.2023.02.002

Abstract

Abstract:

[Objective] This paper summarizes the current status of the paradigm shift in space weather research driven by big data and artificial intelligence, and discusses possible future trends. [Methods] This paper analyzes the challenges faced by the traditional research paradigm and the opportunities brought by artificial intelligence in the space weather big data scenario, investigates the relevant strategies and research progress at home and abroad, conducts a deep analysis of typical cases, and summarizes the current status and characteristics of the paradigm shift in space weather research. [Results] AI technologies have been effectively experimented with in a variety of scientific task types in space weather research, improving scientific efficiency, solving big data scenario difficulties, and spurring a research paradigm shift in the field of space weather. [Limitations] This paper focuses on the summarization of important literature and cases and the analysis of the overall trend of future development. In the future, we plan to further study the status through bibliometrics and other methods, and conduct in-depth discussions on the key issues that need to be addressed in future development. [Conclusions] The research in the field of space weather is undergoing a paradigm shift, whereby the convergence of big data, AI, and domain knowledge is expected to lead to a new paradigm

Key words: space weather, big data, artificial intelligence, research paradigm, domain knowledge, AI-ready

HU Xiaoyan,XU Jiyao,ZOU Ziming. Preliminary Study on Paradigm Shift in Space Weather Research Driven by Big Data and Artificial Intelligence[J]. Frontiers of Data and Computing, 2023, 5(2): 24-36, https://cstr.cn/32002.14.jfdc.CN10-1649/TP.2023.02.002.

Figures/Tables 5

Table 1

Table 2

Fig.1

Fig.2

Table 3

References 32

[1]	邹自明, 胡晓彦, 熊森林. 空间科学大数据的机遇与挑战[J]. 中国科学院院刊, 2018, 33(8): 877-883.
[2]	王赤. 空间科学突破的前瞻和中国的贡献[J]. 中国科学院院刊, 2022, 37(8): 1050-1065.
[3]	王赤, 陈志青, 胡连欢, 等. 我国空间环境天/地基监测平台的发展态势和展望[J]. 航天器环境工程, 2021, 38(3): 225-239.
[4]	吴季, 王赤, 范全林. 中国科学院空间科学战略性先导科技专项实施11年回顾与展望[J]. 中国科学院院刊, 2022, 37(8): 1019-1030.
[5]	程学旗, 梅宏, 赵伟, 等. 数据科学与计算智能: 内涵, 范式与机遇[J]. 中国科学院院刊, 2020, 35(12): 1470-1481.
[6]	NASA. SMD Artificial Intelligence (AI) Initiative[EB/OL]. [2023-02-12]. https://science.nasa.gov/open-sci-ence-overview/smd-ai-initiative.
[7]	NITA G, GEORGOULIS M, KITIASHVILI I, et al. Machine learning in heliophysics and space weather fore-casting: a white paper of findings and recommendations[J]. arXiv preprint arXiv:2006. 12224, 2020.
[8]	ESA. Artificial Intelligence Data Analysis[EB/OL]. [2023-02-12]. https://www.esa.int/Enabling_Support/Preparing_for_the_Future/Discovery_and_Preparation/Artificial_intelligence_in_space.
[9]	中华人民共和国科技部、教育部、工业和信息化部、交通运输部、农业农村部、国家卫生健康委. 关于加快场景创新以人工智能高水平应用促进经济高质量发展的指导意见[EB/OL]. [2023-02-12]. http://www.gov.cn/zhengce/zhengceku/2022-08/12/content_5705154.htm.
[10]	CAMPOREALE E, WING S, JOHNSON J. Machine lea-rning techniques for space weather[M]. Cambridge: Else-vier, 2018.
[11]	BORTNIK J, CAMPOREALE E. Ten ways to apply mac-hine learning in the Earth and Space Sciences[C]// AGU Fall Meeting Abstracts, 2021, 2021: IN12A-06.
[12]	HURLBURT N, CHEUNG M, SCHRIJVER C, et al. Heliophysics Event Knowledgebase for the Solar Dyna-mics Observatory (SDO) and Beyond[J]. Solar Physics, 2012, 1(275): 67-78.
[13]	LI X, WANG Y, LIU R, et al. Reconstructing solar wind inhomogeneous structures from stereoscopic observations in white light: Solar wind transients in 3-D[J]. Journal of Geophysical Research: Space Physics, 2020, 125(7): e2019JA027513.
[14]	PAN Y, JIN M, ZHANG S, et al. TEC map completion using DCGAN and Poisson blending[J]. Space Weather, 2020, 18(5): e2019SW002390.
[15]	LI M, CUI Y, LUO B, et al. Knowledge-Informed Deep Neural Networks for Solar Flare Forecasting[J]. Space Weather, 2022, 20(8): e2021SW002985.
[16]	师立勤, 刘四清, 钟秋珍, 等. 基于机器学习相似度算法的Kp指数预报[J]. 空间科学学报, 2022, 42(2): 199-205.
[17]	MANSHOUR P, BALASIS G, CONSOLINI G, et al. Cau-sality and information transfer between the solar wind and the magnetosphere-ionosphere system[J]. Entropy, 2021, 23(4): 390. doi: 10.3390/e23040390
[18]	YANG Y, SHEN F. Three-Dimensional MHD Modeling of Interplanetary Solar Wind Using Self-Consistent Bou-ndary Condition Obtained from Multiple Observ-ations and Machine Learning[J]. Universe, 2021, 7(10): 371. doi: 10.3390/universe7100371
[19]	SHIMOJO M, HASHIMOTO T, SHIBATA K, et al. Statistical Study of Solar X-ray Jets Observed with the YOHKOH Soft X-ray Telescope[C]// IAU Colloq. 153:Magnetodynamic Phenomena in the Solar Atmosphere-Prototypes of Stellar Magnetic Activity, 1996: 449.
[20]	LAI C, XU J, YUE J, et al. Automatic extraction of gra-vity waves from all-sky airglow image based on machine learning[J]. Remote Sensing, 2019, 11(13): 1516. doi: 10.3390/rs11131516
[21]	BURGER B, MAFFETTONE P M, GUSEV V V, et al. A mobile robotic chemist[J]. Nature, 2020, 583 (7815): 237- 241. doi: 10.1038/s41586-020-2442-2
[22]	TUNYASUVUNAKOOL K, ADLER J, WU Z, et al. Hi-ghly accurate protein structure prediction for the human proteome[J]. Nature, 2021, 596(7873): 590-596. doi: 10.1038/s41586-021-03828-1
[23]	WU T, TEGMARK M. Toward an artificial intelligence physicist for unsupervised learning[J]. Physical Review E, 2019, 100(3): 033311. doi: 10.1103/PhysRevE.100.033311
[24]	孙蒙鸽, 韩涛. 科研智能化与知识服务: 内涵, 实现与机遇[J]. 情报理论与实践, 2021, 44(10): 41.
[25]	XU F, USZKOREIT H, DU Y, et al. Explainable AI: A brief survey on history, research areas, approaches and challenges[C]//Natural Language Processing and Chin-ese Computing: 8th CCF International Conference, NL-PCC 2019, Dunhuang, China, October 9-14, 2019, Pro-ceedings, Part II 8. Springer International Publishing, 2019: 563-574.
[26]	KARNIADAKIS G E, KEVREKIDIS I G, LU L, et al. Ph-ysics-informed machine learning[J]. Nature Reviews Ph-ysics, 2021, 3(6): 422-440.
[27]	XU G, DUONG T D, LI Q, et al. Causality learning: A new perspective for interpretable machine learning[J]. arXiv preprint arXiv:2006.16789, 2020.
[28]	JÖHNK J, WEIßERT M, WYRTKI K. Ready or not, AI comes—an interview study of organizational AI readiness factors[J]. Business & Information Systems Engineering, 2021, 63: 5-20.
[29]	LIANG W, TADESSE G A, HO D, et al. Advances, cha-llenges and opportunities in creating data for trustw-orthy AI[J]. Nature Machine Intelligence, 2022, 4(8): 669-677. doi: 10.1038/s42256-022-00516-1
[30]	MASKEY M, ANSDELL M, COSTES S, et al. NASA Science Mission Directorate Artificial Intelligence Work-shop Update[C]// AGU Fall Meeting Abstracts, 2021: IN11B-01.
[31]	YAO T, WANG J, WAN M, et al. VenusAI: An artificial intelligence platform for scientific discovery on superco-mputers[J]. Journal of Systems Architecture, 2022, 128: 102550. doi: 10.1016/j.sysarc.2022.102550
[32]	UNESCO. Recommendation on Open Science[EB/OL]. [2023-02-12]. https://unesdoc.unesco.org/ark:/48223/pf0000379949.locale=en.

重大任务	数据类型	预期数据规模
子午工程二期（建设中）	太阳射电成像	5类设备，常规模式5分钟1张图（圆环阵1分钟1张图），>90万张图像/年
	地磁探测	秒采样数据，约90套设备，>2.8X10⁹记录/年
	全天空气辉成像	常规模式1分钟1张图，双通道，12个台站，>2.8X10⁶张图片/年
	电离层频高图	常规模式15分钟1张图，15个台站，>50万张图像/年
ASO-S（在轨测试中）	日冕成像	常规模式30秒1张图，>100万张图像/年
	白光全日面成像	常规模式10秒1张图，>300万张图像/年
	莱曼阿尔法全日面成像	常规模式10秒1张图，>300万张图像/年
SMILE（研制中）	紫外极光成像	常规模式1分钟1张图，>50万张图像/年
SMILE（研制中）	软X射线成像	常规模式1分钟1张图，>50万张图像/年

特定科学需求	数据模态	核心问题	模型算法	当前技术指标水平
太阳喷流识别与特征提取	图像数据	分类问题、语义分割问题	Region Growing	准确率>75%
日冕物质抛射事件识别与特征提取	图像数据	分类问题、语义分割问题	Marr-Hildreth+CNN	准确率~70%
极盖区极光弧识别与特征提取	图像数据	目标检测问题	YOLO模型	准确率>80% mIoU>50%
等离子体泡识别与形态提取	图像数据	分类问题、语义分割问题	迁移学习+Resnet18 PSPNet	准确率>97% mIoU>0.88
扩展F事件识别与分类	图像数据	分类问题	DCNN	准确率>93%
电离层行进式扰动事件识别与特征提取	图像数据	目标检测问题、语义分割问题	Faster R-CNN 边缘检测	准确率>95% mIoU>70%
地磁波动事件识别	数值数据	异常检测问题	经验模态分解+小波变换 1D ResNet	准确率>90%
地磁波动频谱分类	数值数据	分类问题	贝叶斯信息准则判据	准确率>93%

	处理原则	效率	主观误差	缺点	适用场景
人工	基于经验和知识	低	大	占用专业人员精力长时间专业人才培养	小规模、典型数据
传统机器	基于先验知识	高	小	先验知识可能有遗漏难以处理复杂型态	大规模、典型数据
人工智能	基于模型学习	高	小	目前大部分模型是黑箱、灰箱	大规模、复杂数据