Automatic identification of Spread-F from Ionograms of Bac Lieu observatory by Using Convolutional Neural Network

Abstract

Spread-F caused by night time ionospheric disturbances is common at equatorial and low-latitudes. Spread-F causes changes in the amplitude and phase of electromagnetic waves passing through the ionosphere and affects the propagation of radio waves in space. Spread-F are classified into 4-types: Frequency Spread-F (FSF), Range Spread-F (RSF), Mixed Spread-F (MSF), and Branch Spread-F (BSF). Accurate identification and classification of Spread-F types helps to statistically analyze the physical characteristics of Spread-F and ionospheric disturbance mechanisms. The detection and classification of Spread-F types from ionograms at Bac Lieu Observatory were carried out based on visual inspection. Currently, modern ionosondes were installed at Bac Lieu Observatory and collect a large amount of ionogram. The manual identification of Spread-F types is very laborious and costly. Therefore, automatic identification and classification of Spread-F is an urgent requirement. In this study, we tested convolutional neural networks to automatically identify and classify Spread-F types from the ionogams of Bac Lieu observatory. The initial results show that the identification and classification of Spread-F types at Bac Lieu have high accuracy, and it is possible to apply artificial intelligence to automatically identify and classify the Spread-F types of this observatory.