Ionospheric structure usually changes dramatically during a strong geomagnetic storm period, which will significantly affect the short-wave communication and satellite navigation systems. It is critically important to make accurate ionospheric predictions under the extreme space weather conditions. However, ionospheric prediction is always a challenge, and pure physical methods often fail to get a satisfactory result since the ionospheric behavior varies greatly with different geomagnetic storms. In this paper, in order to find an effective prediction method, one traditional mathematical method (autoregressive integrated moving average—ARIMA) and two deep learning algorithms (long short-term memory—LSTM and sequence-to-sequence—Seq2Seq) are investigated for the short-term predictions of ionospheric TEC (Total Electron Content) under different geomagnetic storm conditions based on the MIT (Massachusetts Institute of Technology) madrigal observation from 2001 to 2016. Under the extreme condition, the performance limitation of these methods can be found. When the storm is stronger, the effective prediction horizon of the methods will be shorter. The statistical analysis shows that the LSTM can achieve the best prediction accuracy and is robust for the accurate trend prediction of the strong geomagnetic storms. In contrast, ARIMA and Seq2Seq have relatively poor performance for the prediction of the strong geomagnetic storms. This study brings new insights to the deep learning applications in the space weather forecast.

Observations with a global coverage are very important for space physics research and space weather monitoring. However, due to the technical limitations, it would be very expensive or even impossible to achieve a seamless global coverage even with advanced observational devices. It would be useful to fill missing data gaps to create a global map from the available data, but up until now this has been very challenging. Fortunately, the deep learning method, a recent breakthrough in artificial intelligence, may provide an effective way to solve this problem by making full use of data from reliable observations. In this paper, a promising deep learning algorithm, deep convolutional generative adversarial network (DCGAN), is investigated to fill the missing data of total electron content (TEC) map images. The direct use of DCGAN fails to fill missing data for the completion of TEC maps because there are always missing TEC data in some regions, such as oceans, where the features vary with time and geophysical conditions. Thus, no useful information can be utilized by DCGAN to achieve a meaningful image completion. In order to overcome this shortcoming of the original DCGAN method, a novel regularized DCGAN (R‐DCGAN) is proposed by adding an extra discriminator and some widely used reference TEC maps from the International Global Navigation Satellite Systems Service Ionosphere Working Group. The proposed R‐DCGAN method generates satisfactory ionospheric peak structures at different times and geomagnetic conditions, which demonstrate its effectiveness on filling the missing data in TEC maps. The proposed R‐DCGAN framework can be readily extended to a broad application in other fields of space sciences, particularly for addressing the missing observation data issues.

We present an analysis of the responses of quasi 2 day waves (QTDWs) to the 2013 sudden stratosphere warming (SSW) in the mesosphere and lower thermosphere (MLT) region. The study is based on data collected by a meteor radar chain along the 120°E meridian in the Northern Hemisphere which consists of four stations at Mohe (52.5°N, 122.3°E), Beijing (40.3°N, 116.2°E), Wuhan (30.5°N, 114.6°E), and Sanya (18.3°N, 109.6°E). It is the first time that an enhancement of the QTDW in the neutral wind during the 2013 SSW is observed in the midlatitudes in the Northern Hemisphere. During the SSW, the amplification of the QTDW in the low latitudes is the most prominent and the direction of the mean zonal wind shows clear reversion from eastward to westward. The consistent variations of the QTDWs and the mean neutral wind at the four stations are very likely associated with the SSW.

An extensive analysis of quasi 5‐day waves (5DWs) in the mesosphere and lower thermosphere and their responses to a major sudden stratospheric warming (SSW) event (January 2013) are presented. The analysis is based on data conducted from a meteor radar chain in the period from December 2008 to November 2017 and Aura Microwave Limb Sounder satellite. The radar chain includes three stations located at Mohe (MH, 53.5°N, 122.3°E), Beijing (BJ, 40.3°N, 116.2°E), and Wuhan (WH, 30.5°N, 114.6°E). The 5DWs present clear interannual and seasonal variations, which are observed by both the radar wind data and Aura‐temperature data. Interestingly, the ter‐annual oscillation is found to be as important as the commonly recognized annual oscillation and semiannual oscillation at the three stations in both neutral wind components. The 5DWs are strong mainly during the August/September in the meridional component, while they are primarily enhanced in the period of January, April/May, and late summer in the zonal component. An enhancement of 5DWs is observed during the 2013 SSW event in the mesosphere and lower thermosphere region at the three stations. The amplitudes during the SSW are more than 2 times larger than the January average. The strength of the amplification is most prominent at MH and reduces as latitude decreases. According to the results obtained from the radar wind and Aura temperature, the enhanced 5DW modes are W1 at MH, and W2 at BJ and WH. Our results indicate that the amplification of the 5DWs is very likely associated with the 2013 SSW.

We present a study of the quasi 10‐day waves (Q10DWs) in the mesosphere and lower thermosphere (MLT) during a sudden stratospheric warming (SSW) event in February 2018 based on three meteor radars located at Mohe (MH, 53.5°N, 122.3°E), Beijing (BJ, 40.3°N, 116.2°E), and Wuhan (WH, 30.5°N, 116.6°E). The enhanced Q10DWs are observed in the meridional component at MH and in the zonal component at BJ and WH. In the meridional component, the monthly mean Q10DW amplitudes in February 2018 are greater than 10 m/s at MH but weaker than ∼7 m/s at BJ and WH. In the zonal component, monthly mean Q10DWs amplitudes are ∼15 m/s at BJ and WH but only 9 m/s at MH. These latitudinal differences of the Q10DWs in the MLT region are very likely due to the Q10DW activity in the lower atmosphere and the mean background winds. The strong southward winds might bring the wave energy to MH in the meridional component during the occurrence of SSW, while the enhanced Q10DWs in the zonal component at BJ and WH are likely associated with the strong Q10DWs in the lower atmosphere and weakening eastward winds after the central date of the SSW. The significance test indicates that the enhanced Q10DWs in the meridional wind of MH and in the zonal wind of WH are very likely affected by the SSW. In addition, nonlinear interactions between the Q10DWs and other planetary waves also modulate the variations in the MLT region at BJ during the SSW.

We present an analysis of planetary‐scale oscillations during sudden stratospheric warming (SSW) events based on data obtained from a meteor radar located at Mohe (MH, 53.5°N, 122.3°E), the Aura satellite and Modern‐Era Retrospective analysis for Research and Applications, Version 2 data (MERRA2). The planetary‐scale oscillations in the mesosphere and lower thermosphere (MLT) region during eight SSW events from 2012 to 2019 have been statistically investigated. Our analysis reveals that the enhancement or the generation of westward propagating quasi 16‐day oscillation with wavenumber 1 (W1) is a common feature during SSWs over MH. A strong enhancement of the quasi 4‐day oscillation during the 2018/2019 SSW is captured by both radar and satellite observations. The amplified quasi 4‐day oscillation has a period of ~4.3 days in both meridional and zonal winds and with a wavenumber of W2 in the zonal component. Using the meteor radar and MERRA2 data, the vertical structure of the quasi 4‐day oscillation from the stratosphere to the lower thermosphere is derived. The upward propagating feature of the quasi 4‐day oscillation in the meridional component indicates that the oscillation is very likely generated in the lower mesosphere. The mesospheric zonal wind reversal after an elevated stratopause event is observed during the SSW, which results in a negative meridional gradient of the quasi‐geostrophic potential vorticity. Our results not only reveal that the amplified quasi 4‐day oscillation in the MLT region is associated with the 2018/2019 SSW but also suggest that the amplification is originally generated around 60 km due to barotropic/baroclinic instability and propagates upward to MLT region.