Key Points: The Artificial Neural Network Model (ANNM) has good capability of characterizing the variability of the satellite drag‐derived densities The density data during either 1967–2005 or 1967–2013 intervals are used to give a similar long‐term trend from the ANNM The thermospheric density trend relative to the ANNM from 250 to 575 km is about −1.5% to −2.0% per decade without obvious solar flux dependence

The accurate estimation of electron density in the ionosphere is crucial for various applications including remote sensing systems, communication, satellite positioning, and navigation. Previous ionospheric models using artificial neural networks (ANN) are only data‐driven, whose effects are entirely affected by observational data. In this paper, we utilize the results of International Reference Ionosphere (IRI)‐2016 as prior knowledge to develop a low‐latitude (30°N–30°S) three‐dimensional ionospheric electron density model using ANN, namely ANN‐IRI, based on COSMIC radio occultation ionospheric profiles during 2006–2020. The prior knowledge helps ANN‐IRI get a better prediction effect and obtain convergence more quickly. The COSMIC data sets above the altitude 150 km are divided into three sets, a training set (2006–2014 and 2018–2020), a validation set (2016), and a test set (2015 and 2017). For the test set, ANN‐IRI shows a good performance for predicting the electron density, better than ANN (without prior knowledge) and IRI‐2016. And ANN‐IRI behaves better during quiet than disturbed times as well as during low solar activity than high solar activity years. In addition, we corrected effectively the error of ANN‐IRI in the lower ionosphere source from COSMIC data based on IRI‐2016 and spline interpolation. Compared with completely independent data sets from incoherent scatter radars (ISRs) and ROCSAT‐1 in situ observations, the electron density predicted by corrected ANN‐IRI exhibits good similarity. Generally, ANN‐IRI behaves better than ANN and IRI‐2016 for predicting electron density. Our work demonstrates a new possibility of applying deep learning methods with prior knowledge to a broader field of geosciences, particularly for problems of prediction.

Total electron content (TEC) map is one of the important ionospheric parameters. The International Global Navigation Satellite System Service (Ionosphere Working Group) provides the combined vertical TEC maps. However, the postprocessing of the IGS TEC maps may cost quite a long time, and it's not easy for the organization to collect the complete data. It is necessary for researchers to figure out a method to complete the global TEC maps efficiently with regard to the problems of lack of data or not available to the standard IGS TEC. With the rapid development of the deep learning methods, the Deep Convolutional Generative Adversarial Network exhibits the great potential in computer vision. In this paper, we propose a new method called Global and Local GAN (GLGAN) based on the DCGAN and apply it on completing the global TEC maps. Different from the traditional GAN, the GLGAN consists of a generator (or called completion network) and two discriminators. The completion network is powerful enough to Extract features of IGS TEC maps to complete the TEC maps. The design of two discriminators enhances the ability of judging the quality of output images, and improves the accuracy of the completion network. After analyzing the results, we find the GLGAN have a better performance in complicate structures during geomagnetic storm time. The success of the GLGAN in completing the TEC maps suggests that the deep learning methods are able to solve many problems regarding to data and images in ionospheric parameters' reconstruction or forecasting.

One of the most important effects of ionospheric modification by high power, high frequency (HF) waves is the generation of ultra low frequency/extremely low frequency/very low frequency (ULF/ELF/VLF) waves by modulated heating. This paper reviews the scientific achievements of the past five decades regarding the main mechanisms of excitation of ULF/ELF/VLF waves and discusses their characteristics, such as their electrojet dependency, the location of the source region, continuous and discontinuous waves, the number of HF arrays, and the suitable range of the modulation frequency for different proposed mechanisms. Finally, the outlook for future research in this area is presented.