Abstract:Accurate forecasting of F10.7 index is important on short, medium and long-term timescales since F10.7 is an excellent proxy of solar activity and it plays an important role within the Space Weather framework. The analysis of the signatures of transient solar radio emission and its prediction are a challenging task as the underpinning physical processes are typically nonlinear, non-stationary and chaotic. In this paper we want to present three Deep Learning approaches for the daily forecasting of the adjusted F10.7 solar radio flux up to 45 days, using a family of Long Short Term Memory (LSTM) based models. We investigated two novel hybrid architectures: the LSTM model used in combination with Fast Iterative Filtering as decomposition algorithm (FIF-LSTM) and a method based on Multi-Head-Attention architecture (FIF-LSTM-MHA). FIF is a robust decomposition signal method very suitable for analyzing non-linear and non-stationary time series and it is used to separate the original time series into different oscillation components according to frequency, derived without leaving the time domain before to be fed into the neural network. The Attention mechanism is able to keep track of long-term dependencies in data sequences and improve the computational efficiency of the prediction model by reducing the effect of irrelevant information, mimicking human attention and selecting the most critical input. Our comparative analysis evaluated the models’ performance for different time lags and solar activity levels. The results indicated that the hybrid models achieve better performance than the LSTM model for mid-range F10.7 predictions while the LSTM achieves better performance within the first few time lags. FIF-LSTM-MHA gives more promising output for longer forecasts since it tends to smooth the prediction curve due to the peculiarity of the Attention module to discard less relevant features of the time series and highlight the global trend.

Forecast Daily Indices of Solar Activity, F10.7, Using Support Vector Regression Method

Forecasting Medium-Term F10.7 Using the Deep-Learning Informer Model

Study on Optimization of Atmospheric Correction Process Based on 6S Radiative Transfer Model

Deep Learning LSTM-based approaches for 10.7 cm solar radio flux forecasting up to 45-days

F10.7 Daily Forecast Using LSTM Combined With VMD Method

Noise model based ν-support vector regression with its application to short-term wind speed forecasting

On the short-term regional prediction of foF2 based on the support vector machine

The hourly average solar wind velocity prediction based on support vector regression method

The Mid-term Forecast Method of Solar Radiation Index

Solar irradiance forecasting based on RBF neural network

Modeling and Forecasting Ionospheric foF2 Variation Based on CNN-BiLSTM-TPA during Low- and High-Solar Activity Years

Solar irradiance forecasting using multi-layer cloud tracking and numerical weather prediction

Forecasting of solar radiation for a cleaner environment using robust machine learning techniques

Forecasting Stock Composite Index by Fuzzy Support Vector Machines Regression

Short-term solar power prediction using a support vector machine

Solar Radiation Forecast Based on Fuzzy Logic and Neural Networks

Probabilistic Solar Proxy Forecasting with Neural Network Ensembles

Forecasting the Ionospheric F 0 F 2 Parameter One Hour in Advance Using Recurrent Neural Network

A Practical Method to Hourly Forecast the Solar Irradiance

Support vector regression based approach for key index forecasting with applications

A solar radiation intelligent forecasting framework based on feature selection and multivariable fuzzy time series