A method for the automatic detection of solar type III radio bursts with Wind/Waves
Item Type:Conference Paper
Citation:Waters, J. E., Jackman, C. M., Fogg, A. R., Louis, C., Lamy, L., Briand, C., Bonnin, X., Maksimovic, M., Murphy, P. C., Cecconi, B., Issautier, K., Whiter, D. K., A method for the automatic detection of solar type III radio bursts with Wind/Waves, In C. K. Louis, C. M. Jackman, G. Fischer, A. H. Sulaiman, P. Zucca, Dublin Institute for Advanced Studies (Eds.), Planetary, Solar and Heliospheric Radio Emissions IX, 2023. https://doi.org/10.25546/104053
Waters_2023_PRE9_solar_accepted.pdf (Accepted for publication (author's copy) - Peer Reviewed) 2.882Mb
Solar type III radio bursts have a characteristic signature in frequency-time dynamic spectrograms and provide important insight into the dynamics of the Sun and solar wind. Direct physical inferences of the source electrons or their propagation can be made, but radio observations typically contain multiple emission sources which must be accounted for. As well as this, the emission feature of interest must be selected, which can be problematic if done manually, particularly with a large amount of data. The WAVES instrument onboard the Wind spacecraft has accumulated almost thirty years of solar and terrestrial observations, with the majority from L1. Our previous work (Waters et al., 2021b) has focused on the extraction of terrestrial auroral kilometric radiation (AKR) from the Wind/WAVES data, and the removal of noise. In this study we use a similar concept, separating the radio sources by the measured signal variability during a spacecraft spin, but focusing on the parameter space containing interplanetary (IP) type III bursts below 1 MHz. Using an inverse threshold of the spin variability metric and a signal-to-noise ratio (SNR) threshold, the radio data is filtered to return mostly Type III emission above 200 kHz. Integrating the flux density between 500-1045 kHz and choosing a threshold of 1 × 109 W sr−1 to select start times of Type III bursts between 1995 and 2021 yields the detection of 120560 events. Annual event occurrence shows good correlation with yearly-averaged sunspot number across the two solar cycles examined here (P=0.94). Evaluating classification statistics of the selection via visual examination of a random sample of 50 days of Wind data returns a true positive rate of 0.9 and a positive predictive value of 0.4, indicating that the majority of observed Type III bursts are returned. This technique thus provides an initial step to broad statistical examination of the physical properties of Type III bursts using observations from Wind/WAVES or other instruments.
Author: Waters, J. E.; Jackman, C. M.; Fogg, A. R.; Louis, C.; Lamy, L.; Briand, C.; Bonnin, X.; Maksimovic, M.; Murphy, P. C.; Cecconi, B.; Issautier, K.; Whiter, D. K.
Type of material:Conference Paper
Series/Report no:Planetary, Solar and Heliospheric Radio Emissions IX;
Availability:Full text available