Neural Network Model for ESF Occurrence at an Equatorial Station

Equatorial Spread F (ESF) usually seen in the nighttime ionograms as diffused echo is a manifestation of the ionospheric plasma density irregularities. lifting of the F‐region ionization to the higher altitude shortly after local sunset has been identified as the main mechanism responsible for the generation and evolution of equatorial ionospheric plasma density irregularities. Vertical drift velocity () obtained from ionograms can be used as an indicator of the strength of this lifting. Therefore, a neural network model, built around the value and other related parameters for the prediction of ESF occurrence, was trained and validated in this study. The model was a decision tree regressor neural network model and the input parameters were , time of maximum , day of the year and . The target parameter was the ESF occurrence or non‐occurrence (i.e. 1 or 0). Vertical drift velocity values were obtained from available ionograms that were recorded at Ilorin. Attempts were made to vary the number of input parameters in order to see their contribution in predicting ESF occurrence. The best model is the one trained with the four input parameters with 78.62% accuracy during validation. The most revealing is the result obtained when alone was used as input to train the model—in which very poor result was obtained during validation. The results obtained were compared with that of an existing model for ESF prediction.