Automatic Recognition of Low Latitude Ionospheric Irregularities In F Section Based on Deep Learning

doi:10.11871/jfdc.issn.2096-742X.2024.04.011

Abstract

Abstract:

[Objective] Based on the observation data of very high frequency (VHF) coherent scattering radar from the Chinese Meridian Project, the 3m scale ionospheric irregularities in the low-latitude F region are identified and extracted by applying deep learning techniques. [Methods] An intelligent recognition model of ionospheric irregularities is constructed based on the CSPDarknet neural network, and an ionospheric irregularities location model is constructed by using the pre-trained CSPDarknet as the backbone of the Yolo object detection algorithm. [Results] The constructed recognition model can automatically select the low-latitude ionospheric irregularities from the all-day VHF observation data, and extract the height and duration of the irregularities according to the event location model. According to the evaluation of the test set, the F1-score of the recognition model is 85.89%, which improves 5.68% compared to the classical EfficientNet model; the average precision index mAP of the location model can achieve 87.22%, which is 4.32% better than the Yolov5s model. [Limitations] In the process of training the model, we mainly use the observation data from the Hainan Fuke station. To improve the generalization performance of the model, observation data from other stations should be introduced as much as possible. [Conclusions] In this paper, we first propose an intelligent identification and location scheme for ionospheric irregularities by applying advanced deep learning techniques. Compared to traditional threshold methods, the proposed method greatly improves the recognition efficiency, mitigates the dependency on experts, and helps elevate the research efficiency of ionospheric irregularities.

Key words: ionospheric irregularities, VHF radar, CSPDarknet, Yolo algorithm

FANG Shaofeng, ZOU Ziming, HU Xiaoyan. Automatic Recognition of Low Latitude Ionospheric Irregularities In F Section Based on Deep Learning[J]. Frontiers of Data and Computing, 2024, 6(4): 128-138, https://cstr.cn/32002.14.jfdc.CN10-1649/TP.2024.04.011.

Figures/Tables 16

Fig.1

Fig.2

Fig.3

Table 1

Table 2

Fig.4

Fig.5

Fig.6

Fig.7

Fig.8

Fig.9

Table 3

Table 4

Table 5

Fig.10

Fig.11

References 24

[1]	LI G, NING B, OTSUKA Y, et al. Challenges to Equatorial Plasma Bubble and Ionospheric Scintillation Short-Term Forecasting and Future Aspects in East and Southeast Asia[J]. Surv Geophys, 2021, 42: 201-238. https://doi.org/10.1007/s10712-020-09613-5.
[2]	CHEN G, JIN H, YAN J Y, et al. Hainan Coherent Scatter Phased Array Radar (HCOPAR): System Design and Ionospheric Irregularity Observations[J]. IEEE Transactions on Geoscience & Remote Sensing, 2017, (8): 1-9.DOI:10.1109/TGRS.2017.2699280.
[3]	CHEN G, WANG Z, JIN H, et al. A case study of the daytime intense radar backscatter and strong ionospheric scintillation related to the low-latitude E-region irregularities[J/OL]. Journal of Geophysical Research: Space Physics, 2020. DOI: 10.1029/2019JA027532.
[4]	YAN C, CHEN G, WANG Z, et al. Statistical Characteristics of the Low-Latitude E-Region Irregularities Observed by the HCOPAR in South China[J]. Journal of Geophysical Research, A. Space Physics: JGR, 2022(1): 127.
[5]	OKOH D, RABIU B, SHIOKAWA K, et al. First study on the occurrence frequency of equatorial plasma bubbles over West Africa using an all-sky airglow imager and GNSS receivers[J]. Journal of Geophysical Research: Space Physics, 2017, 122(12): 12,430-12, 444.
[6]	LIU K, LI G, NING B, et al. Statistical characteristics of low-latitude ionospheric scintillation over China[J]. Advances in Space Research, 2015, 55(5): 1356-1365.
[7]	NING B Q, HU L H, LI G Z, et al. The first time observations of low-latitude ionospheric irregularities by VHF radar in Hainan[J]. Science China Technological Sciences, 2012, 55: 1189-1197.
[8]	尚社平, 史建魁, 阎敬业, 等. 基于海南VHF雷达观测的低纬E区场向不规则体研究[J]. 空间科学学报, 2014, 34(1): 10. DOI: 10.11728/cjss2014.01.053.
[9]	尚社平, 史建魁, 闫敬业, 等. 海南VHF雷达观测到的低纬F区电离层场向不规则体事件分析[J]. 空间科学学报, 2017, 37(6): 8. DOI: 10.11728/cjss2017. 06.702.
[10]	金晗. 海南地区低纬电离层场向不规则体的观测研究[D]. 武汉大学, 2019.
[11]	MENDOZA M M, CHANG Y C, DMITRIEV A V, et al. Recovery of ionospheric signals using fully convolutional densenet and its challenges[J]. Sensors, 2021, 21(19): 6482.
[12]	殷梦婷, 邹自明, 钟佳. 一种电离层TEC格点预测模型[J]. 空间科学学报, 2021, 41(4): 12.
[13]	SHIDLER S A, RODRIGUES F S. Modeling equatorial ionospheric vertical plasma drifts using machine learning[J]. Earth, Planets and Space, 2020, 72(1): 1-10.
[14]	秦佳媚, 钟鼎坤, 冯学尚, 等. 子午工程数据资源概述[J/OL]. 2021. DOI: 10.11922/csdata.2020.0080.zh.
[15]	金晗, 燕春晓, 陈罡, 等. 2014-2019年海南富克站甚高频相干散射雷达回波数据集[J]. 中国科学数据: 中英文网络版, 2021, 6(2): 10. DOI: 10.11922/csdata.2020.0076.zh.
[16]	JOCHER G, CHAURASIA A, QIU J. 2023. Ultralytics YOLO (Version 8.0.0) [Computer software]. https://github.com/ultralytics/ultralytics.
[17]	BOCHKOVSKIY A, WANG C Y, LIAO H Y M. Yolov4: Optimal speed and accuracy of object detection[J]. arXiv preprint arXiv:2004.10934, 2020.
[18]	HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]// Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.
[19]	WANG C Y, LIAO H Y M, WU Y H, et al. CSPNet: A new backbone that can enhance learning capability of CNN[C]// Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops. 2020: 390-391.
[20]	TAN M, PANG R, LE Q V. Efficientdet: Scalable and efficient object detection[C]// Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 10781-10790.
[21]	SOLOVYEV R, WANG W, GABRUSEVA T. Weighted boxes fusion: Ensembling boxes from different object detection models[J]. Image and Vision Computing, 2021, 107: 104117.
[22]	PASZKE A, GROSS S, MASSA F, et al. Pytorch: An imperative style, high-performance deep learning library[J]. Advances in neural information processing systems, 2019, 32.
[23]	MMPreTrain Contributors. 2023. OpenMMLab’s Pre-training Toolbox and Benchmark (Version 0.15.0) [Computer software]. https://github.com/open-mmlab/mmpretrain.
[24]	MMYolo Contributors. 2018. OpenMMLab Yolo Toolbox and Benchmark [Computer software]. https://github.com/open-mmlab/mmyolo.

	Event	Noise	NoEvent	Unclear
训练集	1197	884	7108	290
验证集	399	294	2369	96
测试集	400	295	2370	97

		Prediction
		Is the Category	Not the Category
Truth	Is the category	True Positive (TP)	False Negative (FN)
Truth	Not the category	False Positive (FP)	True Negative (TN)

Algorithm	Precision	Recall	F1-score
Efficientnet	80.78%	79.64%	80.21%
VGG16	78.16%	81.92%	79.99%
Resnet50	82.55%	80.58%	81.55%
CSPDarknet	86.12%	85.68%	85.89%

Algorithm	Precision	Recall	mAP
Faster Rcnn	68.59%	82.35%	80.37%
Retinanet	66.32%	76.17%	75.94%
Yolov5s	71.16%	87.69%	82.90%
CSPDarknet+BiFPN+WBF	76.83%	87.97%	87.22%