A new method for improving the performance of an ionospheric model developed by multi-instrument measurements based on artificial neural network

There are remarkable ionospheric discrepancies between space-borne (COSMIC) measurements and ground-based (ionosonde) observations, the could decrease accuracies of model developed by multi-source data seriously. To reduce two observational systems, peak frequency (foF2) height (hmF2) derived from COSMIC ionosonde used to develop models an artificial neural network (ANN) method, respectively. The averaged root-mean-square errors (RMSEs) COSPF (COSMIC model), COSPH IONOPF (Ionosonde model) IONOPH 0.58 MHz, 19.59 km, 0.92 MHz 23.40 results indicate that these dependent on universal time, geographic latitude seasons. frequencies measured generally larger than ionosonde’s observations in nighttime or middle-latitudes with amplitude lower 25%, while is smaller polar regions. differences ANN-based maps references show detecting techniques proportional intensity solar radiation. Besides, a new method based ANN technique proposed for improving multiple measurements, RMSEs optimized 14–25% without application method. Furthermore, built more powerful capturing dynamic physics features, such as equatorial ionization, Weddell Sea, mid-latitude summer winter anomalies. In conclusion, significant accuracy physical characteristics observations.