Performance analysis of FFNN and Kriging model on prediction of ionospheric TEC during April 2022 – X 1.1 solar flare

Abstract

Artificial Intelligence (AI) in data analysis has become an integral tool in recent times. In this research, Total Electron Content (TEC) prediction by the Feed Forward Neural Network (FFNN) model is compared with the Ordinary Kriging based Surrogate Model (OKSM) to verify the significance of the sample size required for FFNN and OKSM. To assess the credibility of the constructed models, the FFNN and OKSM prediction models were evaluated on 30 April 2022, during which a Solar Flare (SF) of intensity X 1.1 occurred. The TEC data is taken from Hyderabad (17.31°N and 78.55°E) from IONOLAB data servers. The solar parameters were collected from the NASA OMNIWEB data server. The surrogate model is built to predict the seventh-day TEC values by using the previous six days of TEC data, Whereas the FFNN model uses 146 days of TEC data as training data set to predict the subsequent four days of TEC. The performance of the models is evaluated using statistical parameters like Root Mean Square Error (RMSE), Correlation Coefficient (CC), Mean Absolute Error (MAE) and symmetric Mean Absolute Percentage Error (sMAPE). The results were represented as linear regression scatter plots, showing fewer residuals for the constructed prediction models.

KEYWORDS:

• OKSM; FFNN; GPS; TEC; solar flare

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The research work presented in this paper has been carried out under the Project ID ‘VTU RGS/ DIS-ME/2021-22/5862/1’, funded by VTU, TEQIP, Belagavi, Karnataka.