10:45 AM - 12:15 PM
[PEM09-P02] Prediction of solar cycle variation in solar EUV spectra using multi-frequency radio fluxes
Keywords:space weather, solar radiation, solar activity cycle, machine learning
X-ray (~10 nm) and extreme ultraviolet (EUV: 10 - 124 nm) emissions from the Sun ionizes atoms and molecules in the Earth's upper atmosphere and contribute to the formation of the ionosphere. The ionosphere is used for satellite and terrestrial communications. Since the ionospheric environment fluctuates with the 11-year solar cycle and solar flares, it is necessary to monitor and predict the ionospheric environment to ensure a stable communication environment.
Solar X-rays and EUV emissions that affects the ionosphere are observed using satellites, so observational data is limited. On the other hand, radio emissions from the sun can be observed on the ground. So a radio wave at 2.8 GHz, called F10.7, has been used as a proxy for EUV emission when estimating the impact of solar emission on the ionosphere. However, recent satellite observations of EUV emission spectra have revealed that F10.7 alone cannot explain the actual EUV emissions affecting the ionosphere.
In this study, we first investigate the relationship between radio emission and EUV emission spectra of solar cycle variations using radio data from the Nobeyama Radio Polarimeters (NoRP) and EUV data from the Thermosphere • Ionosphere • Mesosphere • Energetics and Dynamics/the Solar EUV Experiment (TIMED/SEE). NoRP measures the radio flux from the entire Sun at multiple frequencies (1, 2, 3.75, 9.4 GHz), and TIMED/SEE measures EUV emission spectra from 0.5 -190 nm with a resolution of 1 nm. The relationship between these radio emissions and EUV emissions were investigated using daily data for the period 2002-2016, and it was found that radio emissions at all frequencies correlate well with EUV emissions. In particular, radio emissions at lower frequencies tend to have a better correlation with EUV emissions. In addition, the correlation became worse in the EUV channels which contain more line emission from the chromosphere, and the slope of the correlation also appeared to change.
Because the emission mechanisms of radio and EUV emission are different, it is difficult to accurately derive the relationship between these two types of emissions only from the comparison of observed data. Therefore, in this study, we try to reproduce a specific solar EUV emission spectrum from multiple frequencies of radio observation data using an artificial neural network (ANN) with reference to Zhang & Paxton (2018). The input radio data were not only the NoRP data mentioned above, but also the Australian Learmonth solar radio telescopes monitor data used in Zhang & Paxton (2018) (245, 440, 610, 1415, 2695, 4995, 8800, and 15400 MHz) were also used. Using daily observation data between 2002 and 2016, we reproduced the solar EUV emission spectra obtained with TIMED/SEE using ANN with correlation coefficients better than 0.90 for most EUV wavelength. The frequencies of radio waves that particularly contribute to the reproduction of EUV emission was investigated, and it was found that the frequencies with high contribution vary with wavelength, with 2 GHz having a high contribution in the short wavelength (10 - 50 nm), and 1 and 2 GHz in the longer wavelength (50 - 124 nm).
In this presentation, we will report these results and discuss how the contribution of each frequency varies with the solar activity.
Solar X-rays and EUV emissions that affects the ionosphere are observed using satellites, so observational data is limited. On the other hand, radio emissions from the sun can be observed on the ground. So a radio wave at 2.8 GHz, called F10.7, has been used as a proxy for EUV emission when estimating the impact of solar emission on the ionosphere. However, recent satellite observations of EUV emission spectra have revealed that F10.7 alone cannot explain the actual EUV emissions affecting the ionosphere.
In this study, we first investigate the relationship between radio emission and EUV emission spectra of solar cycle variations using radio data from the Nobeyama Radio Polarimeters (NoRP) and EUV data from the Thermosphere • Ionosphere • Mesosphere • Energetics and Dynamics/the Solar EUV Experiment (TIMED/SEE). NoRP measures the radio flux from the entire Sun at multiple frequencies (1, 2, 3.75, 9.4 GHz), and TIMED/SEE measures EUV emission spectra from 0.5 -190 nm with a resolution of 1 nm. The relationship between these radio emissions and EUV emissions were investigated using daily data for the period 2002-2016, and it was found that radio emissions at all frequencies correlate well with EUV emissions. In particular, radio emissions at lower frequencies tend to have a better correlation with EUV emissions. In addition, the correlation became worse in the EUV channels which contain more line emission from the chromosphere, and the slope of the correlation also appeared to change.
Because the emission mechanisms of radio and EUV emission are different, it is difficult to accurately derive the relationship between these two types of emissions only from the comparison of observed data. Therefore, in this study, we try to reproduce a specific solar EUV emission spectrum from multiple frequencies of radio observation data using an artificial neural network (ANN) with reference to Zhang & Paxton (2018). The input radio data were not only the NoRP data mentioned above, but also the Australian Learmonth solar radio telescopes monitor data used in Zhang & Paxton (2018) (245, 440, 610, 1415, 2695, 4995, 8800, and 15400 MHz) were also used. Using daily observation data between 2002 and 2016, we reproduced the solar EUV emission spectra obtained with TIMED/SEE using ANN with correlation coefficients better than 0.90 for most EUV wavelength. The frequencies of radio waves that particularly contribute to the reproduction of EUV emission was investigated, and it was found that the frequencies with high contribution vary with wavelength, with 2 GHz having a high contribution in the short wavelength (10 - 50 nm), and 1 and 2 GHz in the longer wavelength (50 - 124 nm).
In this presentation, we will report these results and discuss how the contribution of each frequency varies with the solar activity.