Deep temporal convolutional networks for F10.7 radiation flux short-term forecasting

Abstract

Abstract. F10.7, the solar flux at a wavelength of 10.7 cm (F10.7), is often used as an important parameter input in various space weather models and is also a key parameter for measuring the strength of solar activity levels. Therefore, it is valuable to study and forecast F10.7. In this paper, the temporal convolutional network (TCN) approach in deep learning is used to predict the daily value of F10.7. The F10.7 series from 1957 to 2019 are used. The data during 1957–1995 are adopted as the training dataset, the data during 1996–2008 (solar cycle 23) are adopted as the validation dataset, and the data during 2009–2019 (solar cycle 24) are adopted as the test dataset. The leave-one-out method is used to group the dataset for multiple validations. The prediction results for 1–3 d ahead during solar cycle 24 have a high correlation coefficient (R) of 0.98 and a root mean square error (RMSE) of only 5.04–5.18 sfu. The overall accuracy of the TCN forecasts is better than the autoregressive (AR) model (it only takes past values of the F10.7 index as inputs) and the results of the US Space Weather Prediction Center (SWPC) forecasts, especially for 2 and 3 d ahead. In addition, the TCN model is slightly better than other neural network models like the backpropagation (BP) neural network and long short-term memory (LSTM) network in terms of the solar radiation flux F10.7 forecast. The TCN model predicted F10.7 with a lower root mean square error, a higher correlation coefficient, and a better overall model prediction.