Prediction of foF2 using Neural Network at Thailand equatorial latitude station, Chumphon

This paper proposes the development of a Neural Network (NN) model for the prediction of the F2 layer critical frequency (foF2) at the magnetic equator over Chumphon (10.72°N, 99.37°E, dip angle 3.3°N), Thailand and then compared with the IRI model and the experimental ones. The feed forward network with backpropagation algorithm has been developed for predicting the foF2 values. The NN is trained with the daily hourly values of foF2 during the period from 2004 to 2008 and the input parameters affecting the foF2 variability including the hour number, day number, F10.7 index and sunspot number (SSN). To examine the performance of the proposed NN, the root mean square error (RMSE) of the observed foF2, the proposed NN model and the IRI (both CCIR and URSI options) model are compared in 2009. The results show that the NN model predicts the foF2 values close to the observed data, particularly during daytime. Moreover, the NN model can predicts more accurate than the IRI model that is supported by the lower RMSE. However, the NN model provides slightly deviation of prediction during pre-sunrise hours since the observed foF2 data for NN training in this periods are fewer than those during daytime.