Solar radio spikes most normal observational functions are their short period and narrow bandwidth. The 2 left columns reveal the labeled spikes in datasets, and the ideal two columns are found spikes by the enhanced network. In order to evaluate the distribution of spikes, we plot pie charts of their periods, bandwidths, and relative bandwidths, as revealed in Figure 2. At the metric wavelengths, the periods likewise reveal an outstanding symmetrical distribution, and many of the spikes are within 40-50 ms. The values are not similarly distributed for the bandwidths and the relative bandwidths. The mean period (11.2 ms) and bandwidth (21.8 MHz) of no quantifiable frequency-drift spikes are slightly smaller than the ways of positive (12.1 ms and 25.1 MHz) and negative (12.2 ms and 23.7 MHz) frequency-drift spikes.
Figure 1. (a) Precision-Recall relationship. (b) In comparing labeling and discovering, the left column is 2 images with identified spikes, and the right is images with discovered points from the better network.
The relationship in between Precision and Recall is displayed in Figure 1( a). Precision is TP/( TP+FP), representing what percentage of the target identified by the model is real, that is, the real spikes. Amongst them, the Recall is equivalent to TP/( TP+FN), which implies what percentage of the real target can be found by the design. In the formula, TP, fn, and fp are the spikes numbers with favorable predictions and labels, favorable predictions but negative labels, and negative forecasts but favorable labels, respectively. In Figure 1( a), when Recall is 0.64, Precision equals 0.85. As soon as discover the inclined bounding frames of spikes exist, it is required to assess and evaluate the efficiency of the detection. AP value is used as an assessment sign. Its figured out by the Precision-Recall curve and is equal to the area enclosed by the curve. The datasets are input into the better and initial YOLOv5s network2. The models are trained according to the main pre-training weights file of YOLOv5s.pt3 with a training time of 1200. The detection results from these two models are compared.
Figure 1( b) shows a comparison of the detection results. The figure includes four sub-images. The two left columns reveal the labeled spikes in datasets, and the ideal two columns are discovered spikes by the improved network. The enhanced network carried out well concerning little targets detection precision and likely bounding frames regression precision.
The YOLOv5s network is improved concerning the common attributes of spikes. In the enhanced network, likely bounding frames are added to detect their frequencies drifting with time. Offered numerous morphologies, short periods, and narrow bandwidths, attention and feature combination system modules are added. The meter-wavelength and decimeter-wavelength spikes observed by the SBRS/Huairou and the spectrograph in the CSO are used to conduct experiments, respectively. The outcomes show that the AP worth obtained by the better network is 74%. The value is nearly 14% higher than that of the initial network.
The better network finds 9709 (1379) decimeter- (meter-) wavelength spikes in two events that took place on 2005 January 20 (2016 July 18) with periods, bandwidths, relative bandwidths, and frequency-drift rates. According to the frequency varieties and frequency-drift instructions, the spikes are classified.
Solar radio spikes most normal observational functions are their brief period and narrow bandwidth. They appear on the solar radio dynamic spectrogram as a great deal of narrow-band type-III bursts, spikes, dots, sub-second clumps, groups, chains, and other narrow-band structures from decimeter to decameter wavelengths (Feng et al.( 2018 ), Tan et al.( 2019 )). We have improved the YOLOv5s network design for these attributes by including inclined bounding frames and attention and function fusion system modules. Recently, Hou et al. (2020) determined and drawn out the solar radio spikes using the Faster Region-based Convolutional Neutral Network (Faster R-CNN), providing the AP worth 91%.
The decimeter- and meter-wavelength spikes observed by the Solar Broad-band Radio Spectrometer in Huairou and the Chashan Solar Radio Observatory spectrograph are used to conduct experiments, respectively. These analytical outcomes and findings constrain solar radio spikes development.
Histograms of spikes at decimeter (the left column) and meter (the right column) wavelengths. The very first, 2nd, and third rows are the period, the bandwidth, and the relative bandwidth distributions, respectively.
In order to analyze the distribution of spikes, we plot histograms of their periods, bandwidths, and relative bandwidths, as revealed in Figure 2. At the metric wavelengths, the durations also show an excellent balanced distribution, and most of the spikes are within 40-50 ms. The worths are not similarly dispersed for the bandwidths and the relative bandwidths. We bring out a statistical research study on the categorized spikes and discover the following main outcomes:
( 1) The period declines with the boost of frequency, the bandwidth increases with frequency and decreases with period, and the relative bandwidth remains unchanged with frequency and period. The duration (bandwidth) at the decimeter wavelengths is about one-quarter (4-5 times) of that at the meter wavelengths. The decimeter (meter) wavelength period has an exceptional in proportion circulation and mainly depends on 11-12( 40-50) ms. The bandwidth and relative bandwidth at the two wavelengths show unbalanced circulations.
( 2) At the decimeter wavelengths, spikes with no measurable frequency-drift rates have the most substantial number, the negative is the 2nd, and the positive is the last. The mean period (11.2 ms) and bandwidth (21.8 MHz) of no quantifiable frequency-drift spikes are slightly smaller than the ways of positive (12.1 ms and 25.1 MHz) and unfavorable (12.2 ms and 23.7 MHz) frequency-drift spikes. At the meter wavelengths, spikes with unfavorable and favorable frequency-drift rates are nearly similar, and no measurable frequency-drift rates are the least.
Based on the recent paper by Lv, P.R., Hou, Y.C., Feng, S.W. et al.. Automated detection and analytical study of solar radio spikes, Astrophys Space Sci 368, 14 (2023 ). DOI: 10.1007/ s10509-023-04172-8.
Recommendations.
Feng, S.W., Chen, Y., Li, C.Y. et al.. Harmonics of Solar Radio Spikes at Metric Wavelengths. Sol Phys 293, 39 (2018 ).
Hou, Y.C., Zhang, Q.M., Feng, S.W. et al.. Recognition and Extraction of Solar Radio Spikes Based on Deep Learning. Sol Phys 295, 146 (2020 ).
Tan B, Chen N, Yang Y H, et al. Solar Fast-drifting Radio Bursts in an X1.3 Flare on 2014 April 25 [J] The Astrophysical Journal, 2019, 885( 1 ):90.
* Full list of authors: P.R. Lv, Y.C. Hou, S.W. Feng, Q.F. Du and C.M. Tan.
