foF2 Estimation with TEC Observation over China Based on Random Forest Method

The critical frequency of F2 layer (foF2) plays an important role in high-frequency communication, but the ionosonde to obtain foF2 measurements is expensive which leads to scarce stations and measurements. However, the Global Navigation Satellite System (GNSS) receiver to obtain total electron content (TEC) measurements are inexpensive and the observation sites are dense. Due to a strong correlation between the parameters of foF2 and TEC, this paper proposes a random forest (RF) method for predicting foF2 using TEC. A regional foF2 prediction model was constructed using foF2 and TEC data of 10 stations in China. Among them, 8 stations data are used for training the RF model, and the remaining 2 station data are used for model accuracy evaluation. The results show that the RMSE of RF model is 0.77MHz in mid-latitude region and 1.09MHz in low-latitude region, and the prediction accuracy is better than IRI2016 model.