基于最大似然估计的远紫外遥感反演电离层电子密度算法

冯桃君; 于钱; 张凯

doi:10.11728/cjss2022.06.211115118

基于最大似然估计的远紫外遥感反演电离层电子密度算法

doi: 10.11728/cjss2022.06.211115118

北京卫星环境工程研究所　北京　100094

基金项目: 国家重点研发计划项目资助（2016 YFB0501300，2016 YFB0501304）

详细信息

作者简介:
冯桃君：E-mail：taozi_1227@126.com

中图分类号: P356
计量
- 文章访问数: 55
- HTML全文浏览量: 12
- PDF下载量: 18
- 被引次数: 0
出版历程
- 收稿日期: 2021-11-15
- 录用日期: 2021-04-15
- 修回日期: 2022-05-05
- 网络出版日期: 2022-11-09

An Algorithm for Retrieving Ionospheric Electron Density from Far Ultraviolet Remote Sensing Based on Maximum Likelihood Estimation

Beijing Institute of Spacecraft Environment Engineering, Beijing 100094

摘要

摘要: 原子氧135.6 nm夜气辉主要由氧离子O⁺与电子的辐射复合反应生成，一些星载远紫外遥感观测任务证实135.6 nm夜气辉可用于反演电离层电子密度。针对远紫外临边遥感观测反演电离层电子密度，分析了135.6 nm夜气辉辐射强度与电子密度之间的非线型前向模型，基于离散反演理论设计了从夜间135.6 nm临边观测数据反演电子密度高度分布的反演算法，算法应用最大似然估计通过迭代求解电离层参数的最佳拟合值。通过仿真计算了TIMED卫星上全球紫外成像仪GUVI观测的反演结果，验证了本反演算法的可行性。对GUVI的实际观测数据进行反演，获得了电子密度高度分布。通过与GUVI数据的电离层参数对比分析得出，本文建立的反演模型使N_mF₂被高估，同时使h_mF₂被低估。对于不同的太阳活动强度，N_mF₂和 h_mF₂的系统误差分别在10%和5%以内，能较精确地获得电离层参数。精确获得电离层电子密度信息对于提高空间天气预报及电离层模型的修正具有重要意义。
- 远紫外 /
- 遥感 /
- 电离层 /
- 反演算法 /
- 电子密度
Abstract: The OI 135.6 nm nighttime emission is dominantly produced by radiative recombination of O⁺ ions and electrons. Many previous space-based Far Ultraviolet (FUV) remote sensing experiments have demonstrated that OI 135.6 nm nighttime intensity can be used to infer the ionospheric F region electron density. This paper firstly presents a forward model specifying the nonlinear relationship between electron density and 135.6 nm nightglow intensity. Then, we develop an algorithm to infer the altitude profile of electron density from the nighttime 135.6 nm limb intensity measurements using Discrete Inverse Theory (DIT). The algorithm applies maximum likelihood method to iteratively seek the most probable values of the ionospheric parameters. The viability of this algorithm is verified through performing the simulation of the synthetic 135.6 nm limb observation data generated from forward model using the TIMED/GUVI limb scan configuration. Finally, we invert the realistic GUVI limb observation measurements and obtain the retrieved Electron Density Profile (EDP). The comparison between retrieved ionospheric parameters and GUVI products suggests that the forward model tends to overestimate the N_mF₂ and underestimate the h_mF₂. The systematic error is within 10% for N_mF₂ and 5% for h_mF₂ for different level of solar activity. Determining ionosphere electron density with high precision could help improve the ionospheric model and forecast the space weather.
- Far ultraviolet /
- Remote sensing /
- Ionosphere /
- Inversion algorithm /
- Electron density

HTML全文

图 1 GUVI的扫描示意

Figure 1. Schematic of GUVI’s scan imaging

下载: 全尺寸图片幻灯片

图 2 10次反演结果及反演结果均值与电离层参数真值对应的电子密度高度（a）和 135.6 nm 辐射强度随视线切点高度的分布（b）。10次反演叠加的噪声相同但迭代初值不同

Figure 2. (a) Electron density altitude profiles and (b) 135.6 nm radiation intensity against tangent height of line of sight corresponding to the ten retrieved solutions, mean of solutions, true values of ionospheric parameters. Ten inversion runs have identical noise but different initial guess

下载: 全尺寸图片幻灯片

图 3 $ {\chi }^{2} $ 随迭代步数的变化（不同曲线代表不同的反演，每次反演的迭代初值和噪声不同）

Figure 3. Variation of $ {\chi }^{2} $ with number of iterations (Different curve represents different inversion run. Each inversion run is different in both iteration initial value and added errors)

下载: 全尺寸图片幻灯片

图 4 100次反演的结果分布及百分比误差的概率密度函数 (PDF)统计直方图（每次反演的迭代初值和噪声不同）

Figure 4. Distributions of three ionospheric parameters with respect to true values and the Probability Density Function (PDF) histograms of percentage differences for 100 inversion runs (Each inversion run is different in both iteration initial value and added errors)

下载: 全尺寸图片幻灯片

图 5 100次反演结果及反演结果均值与电离层参数真值对应的电子密度高度（a）和135.6 nm辐射强度随视线切点高度分布（b）（每次反演的迭代初值与噪声不同)

Figure 5. (a) Electron density altitude profiles and (b) 135.6 nm radiation intensity against tangent height of line of sight corresponding to the one hundred retrieved solutions, mean of solutions, true values of ionospheric parameters (Each inversion run is different in both iteration initial value and added errors)

下载: 全尺寸图片幻灯片

图 6 三种仪器响应度情况下的N_mF₂和 h_mF₂反演百分比误差对比

Figure 6. Percentage difference of N_mF₂ and h_mF₂ for three responsivity levels

下载: 全尺寸图片幻灯片

图 7 2002年7月20日（a）和2007年10月4日（b）GUVI 135.6 nm临边观测反演结果典型示例及与相同条件下IRI2016模型与GUVI数据的EDP及基于IRI2016模型仿真反演结果的对比

Figure 7. Comparison among the typical inversion results from GUVI 135.6 nm observations, IRI2016 EDP, GUVI EDP product and the inversion results from simulated data based on IRI2016 model on 20 July 2002 (a) and 4 October 2007 (b) under the same condition

下载: 全尺寸图片幻灯片

图 8 2002年7月20日的观测数据反演获得的电离层参数与GUVI数据电离层参数的比较

Figure 8. Comparison between retrieved ionospheric parameters from observations and GUVI data on 20 July 2002

下载: 全尺寸图片幻灯片

图 9 2007年10月4日的观测数据反演获得的电离层参数与GUVI数据电离层参数的比较

Figure 9. Comparison between retrieved ionospheric parameters from observations and GUVI data on 4 October 2007

下载: 全尺寸图片幻灯片

表 1 3种响应度情况的反演误差(%)数据对比

Table 1. Comparison of the retrieved errors (%) of three responsivity cases

Ionospheric parameter	0.1			1			10
Ionospheric parameter	max	min	RMS	max	min	RMS	max	min	RMS
N_mF₂	39.98	0.48	12.04	23.28	0.04	6.33	5.19	0.05	1.85
h_mF₂	32.94	0.01	7.89	9.46	0.03	2.48	2.90	0.01	0.85

下载: 导出CSV

参考文献(21)

[1]	BRUNINI C, AZPILICUETA F, JANCHES D. An attempt to establish a statistical model of the day-to-day variability of the N_mF₂ and h_mF₂ parameters computed from IRI[J]. Advances in Space Research, 2015, 55(8): 2033-2040 doi: 10.1016/j.asr.2014.07.023
[2]	COOK K, FONG C J, WENKEL M J, et al. FORMOSAT-7/COSMIC-2 GNSS radio occultation constellation mission for global weather monitoring[C]//Proceedings of 2013 IEEE Aerospace Conference. Big Sky: IEEE, 2013: 1-8
[3]	TSAI L C, SU S Y, LIU C H. Global morphology of ionospheric F-layer scintillations using FS3/COSMIC GPS radio occultation data[J]. GPS Solutions, 2017, 21(3): 1037-1048 doi: 10.1007/s10291-016-0591-4
[4]	MEIER R R. Ultraviolet spectroscopy and remote sensing of the upper atmosphere[J]. Space Science Reviews, 1991, 58(1): 1-185 doi: 10.1007/BF01206000
[5]	HSU M L, RAJESH P K, LIU J Y, et al. Ionospheric electron density concurrently derived by TIP and GOX of FORMOSAT-3/COSMIC[J]. Terrestrial, Atmospheric and Oceanic Sciences, 2009, 20(1): 207-214 doi: 10.3319/TAO.2008.04.24.02(F3C)
[6]	RAJESH P K, LIU J Y, HSU M L, et al. Ionospheric electron content and N_mF₂ from nighttime OI 135.6 nm intensity[J]. Journal of Geophysical Research, 2011, 116(A2): A02313
[7]	DYMOND K F, NICHOLAS A C, BUDZIEN S A, et al. The special sensor ultraviolet limb imager instruments[J]. Journal of Geophysical Research, 2017, 122(2): 2674-2685 doi: 10.1002/2016JA022763
[8]	MENDE S B, HEETDERKS H, FREY H U, et al. Far ultraviolet imaging from the image spacecraft. 1. System design[J]. Space Science Reviews, 2000, 91(1/2): 243-270 doi: 10.1023/A:1005271728567
[9]	PAXTON L J, CHRISTENSEN A B, MORRISON D, et al. GUVI: a hyperspectral imager for geospace[C]//Proceedings of SPIE 5660, Instruments, Science, and Methods for Geospace and Planetary Remote Sensing. Honolulu: SPIE, 2004: 228-240
[10]	EASTES R W, MCCLINTOCK W E, BURNS A G, et al. The global-scale observations of the limb and disk (GOLD) mission[J]. Space Science Reviews, 2017, 212(1/2): 383-408
[11]	DEMAJISTRE R, PAXTON L J, MORRISON D, et al. Retrievals of nighttime electron density from Thermosphere Ionosphere Mesosphere Energetics and Dynamics (TIMED) mission Global Ultraviolet Imager (GUVI) measurements[J]. Journal of Geophysical Research, 2004, 109(A5): A05305
[12]	DYMOND K F, BUDZIEN S A, CHUA D H, et al. Tomographic reconstruction of the low-latitude nighttime electron density using FORMOSAT-3/COMSIC radio occultation and UV photometer data[J]. Terrestrial, Atmospheric and Oceanic Sciences, 2009, 20(1): 215-226 doi: 10.3319/TAO.2008.01.15.01(F3C)
[13]	QIN J Q, MAKELA J J, KAMALABADI F, et al. Radiative transfer modeling of the OI 135.6 nm emission in the nighttime ionosphere[J]. Journal of Geophysical Research, 2015, 120(11): 10116-10135 doi: 10.1002/2015JA021687
[14]	DYMOND K F, BUDZEN S A, COKER C, et al. The tiny ionospheric photometer (TIP) on the constellation observing system for meteorology, ionosphere, and climate (COSMIC/FORMOSAT-3)[J]. Journal of Geophysical Research, 2016, 121(10): 10614-10622
[15]	DYMOND K F, NEE J B, THOMAS R J. The tiny ionospheric photometer: an instrument for measuring ionospheric gradients for the COSMIC constellation[J]. Terrestrial, Atmospheric and Oceanic Sciences, 2000, 11(1): 273-290 doi: 10.3319/TAO.2000.11.1.273(COSMIC)
[16]	MENKE W. Geophysical Data Analysis: Discrete Inverse Theory[M]. San Diego: Academic Press, 1989
[17]	MEIER R R, PICONE J M. Retrieval of absolute thermospheric concentrations from the far UV dayglow: an application of discrete inverse theory[J]. Journal of Geophysical Research, 1994, 99(A4): 6307-6320 doi: 10.1029/93JA02775
[18]	PUJOL J. The solution of nonlinear inverse problems and the Levenberg-Marquardt method[J]. Geophysics, 2007, 72(4): W1-W16 doi: 10.1190/1.2732552
[19]	BELLAVIA S, GRATTON S, RICCIETTI E. A Levenberg–Marquardt method for large nonlinear least-squares problems with dynamic accuracy in functions and gradients[J]. Numerische Mathematik, 2018, 140(3): 791-825 doi: 10.1007/s00211-018-0977-z
[20]	HUMM D C, PAXTON L J, CHRISTENSEN A B, et al. Design and performance of the global ultraviolet imager (GUVI)[C]//Proceedings of SPIE 3445, EUV, X-Ray, and Gamma-Ray Instrumentation for Astronomy IX. San Diego: SPIE, 1998: 2-12
[21]	高超, 刘黎军, 王志强, 等. “风云三号”电离层掩星气候学数据处理方法[J]. 航天器环境工程, 2021, 38(3): 311-317 GAO Chao, LIU Lijun, WANG Zhiqiang, et al. Processing climatological data observed from FY-3 ionospheric occultation[J]. Spacecraft Environment Engineering, 2021, 38(3): 311-317