Yuqiang Zhang

In this study, the temporal and spatial distribution characteristics of ionospheric scintillation in the East Asian sector are statistically analyzed based on S4 data provided by the COSMIC-1 occultation dataset and solar–terrestrial spatial environment parameters from 2007 to 2018. The results show that scintillation activity has an obvious distribution pattern with local time: the frequency gradually increases from 17:00 in the evening, with the peak concentrated at 22:00–01:00 at night; in terms of seasonal variation, scintillation activity is highest in spring and fall, followed by summer, and lowest in winter; and, regarding annual variation, it is highly correlated with the solar activity. Further analyses show that scintillation activity is strongly correlated with geomagnetic activity. On this basis, this study constructs a two-layer LSTM deep learning model based on weighted regression to realize S4 numerical forecasting for the next 1 h in the middle- and low-latitude regions of China, using F10.7, Kp, Dst, sunspot number, solar wind vertical velocity, and historical S4 values as inputs. The model demonstrates robust predictive performance on the validation dataset containing 8760 samples, with a mean squared error of 0.00546 and an absolute error that is distributed within the interval [−0.2, 0.2] 98% of the time, indicating strong accuracy and robustness. These results suggest that the proposed model provides a high-precision tool for ionospheric scintillation warning.

All-sky meteor radars are widely employed to observe the near-space atmospheric wind field, a crucial parameter of the near-space environment. Owing to the spatiotemporal uncertainty in meteor count distribution, meteor radars may encounter measurement errors and data gaps when retrieving atmospheric wind fields. Using Kalman filter assimilation technology in combination with the HWM14, this study utilizes atmospheric wind field observation data from the Qujing meteor radar (25.6°N, 103.7°E) to study atmospheric horizontal wind fields within the altitude range of 80–100 km. The assimilation results indicate that the accuracy of the HWM14′s atmospheric wind field is significantly improved after Kalman filter-based assimilation. The discrepancy between the assimilated wind field analysis values and the meteor radar wind field values is notably reduced: the average maximum error of zonal wind speed is 12.0 m/s at 90 km altitude, representing a 55.0% improvement compared to the pre-assimilation state; the average maximum error of the meridional wind speed is 14.2 m/s at 100 km altitude, a 53.4% improvement. Furthermore, the standard deviation of the deviation between the assimilated wind field analysis values and the meteor radar wind field values is also substantially decreased. The assimilated atmospheric wind field information holds great significance for further investigating atmospheric disturbance variations and dynamic processes in the near-space region.

Accurately predicting evaporation duct height (EDH) is a crucial technology for enabling over-the-horizon communication and radar detection at sea. To address the issues of overfitting in neural network training and the low efficiency of manual hyperparameter tuning in conventional evaporation duct height (EDH) prediction, this study proposes the application of Bayesian optimization (BO)-based deep learning techniques to EDH forecasting. Specifically, we developed a novel BO–LSTM hybrid model to enhance the predictive accuracy of EDH. First, based on the CFSv2 reanalysis data from 2011 to 2020, we employed the NPS model to calculate the hourly evaporation duct height (EDH) over the Yongshu Reef region in the South China Sea. Then, the Mann–Kendall (M–K) method and the Augmented Dickey–Fuller (ADF) test were employed to analyze the overall trend and stationarity of the EDH time series in the Yongshu Reef area. The results indicate a significant declining trend in EDH in recent years, and the time series is stationary. This suggests that the data can enhance the convergence speed and prediction stability of neural network models. Finally, the BO–LSTM model was utilized for 24 h short-term forecasting of the EDH time series. The results demonstrate that BO–LSTM can effectively predict EDH values for the next 24 h, with the prediction accuracy gradually decreasing as the forecast horizon extends. Specifically, the 1 h forecast achieves a root mean square error (RMSE) of 0.592 m, a mean absolute error (MAE) of 0.407 m, and a model goodness-of-fit (R²) of 0.961. In contrast, the 24 h forecast shows an RMSE of 2.393 m, MAE of 1.808 m, and R² of only 0.362. A comparative analysis between BO–LSTM and LSTM reveals that BO–LSTM exhibits marginally superior accuracy over LSTM for 1–15 h forecasts, with its performance advantage becoming increasingly pronounced for longer forecast horizons. This confirms that the Bayesian optimization-based hyperparameter tuning method significantly enhances model prediction accuracy.

A numerical simulation of Perkins instability in the midlatitude F-region ionosphere is developed in this study. The growth of nighttime plasma density perturbation excited by Perkins instability was successfully reproduced. The simulated results show that the ionospheric perturbation structure elongated from northwest (NW) to southeast (SE) was generated from initial random seeding by applying a very large southeastward neutral wind (200 m/s). The domain wave vector direction agreed with the linear Perkins theory. Our simulated results were consistent with the previous observations and simulations. To investigate the influence of background ionospheric multi-factors on the generation of nighttime medium-scale traveling ionospheric disturbance (MSTID), we simulated the evolution process of ionospheric perturbations under initial background ionospheric conditions. The simulated results indicate the importance of neutral scale height on the development of nighttime MSTID and suggest that a smaller neutral scale height would amplify the amplitude of ionospheric perturbations. The influences of gravity wave (GW) activity and polarized electric field seeding from plasma instability in the E region are also discussed in this study. We conclude that the additional seeding processes play a major role in the accelerated Perkins instability and amplify ionospheric perturbations. The electrodynamic coupling process has a greatly significant effect on the growth rate of Perkins instability compared to GW activity.

Developing short-term forecasting models for global atmospheric temperature and wind in near space is crucial for understanding atmospheric dynamics and supporting human activities in this region. While numerical models have been extensively developed, deep learning techniques have recently shown promise in improving atmospheric forecasting accuracy. In this study, convolutional long short-term memory (ConvLSTM) and convolutional gated recurrent unit (ConvGRU) neural networks were applied to build for short-term global-scale forecasting model of atmospheric temperature and wind in near space based on the MERRA-2 reanalysis dataset from 2010–2022. The model results showed that the ConvGRU model outperforms the ConvLSTM model in the short-term forecast results. The ConvGRU model achieved a root mean square error in the first three hours of approximately 1.8 K for temperature predictions, and errors of 4.2 m/s and 3.8 m/s for eastward and northward wind predictions on all 72 isobaric surfaces. Specifically, at a higher altitude (on the 1.65 Pa isobaric surface, approximately 70 km above sea level), the ConvGRU model achieved a RMSE of about 2.85 K for temperature predictions, and 5.67 m/s and 5.17 m/s for eastward and northward wind. This finding is significantly meaningful for short-term temperature and wind forecasts in near space and for exploring the physical mechanisms related to temperature and wind variations in this region.

Meteor radar is one of the key tools for studying the atmospheric dynamics in the mesosphere and lower thermosphere. The physical parameters obtained by meteor radar inversion can provide important statistical information for research. The daily and annual variations in meteor azimuth distribution detected by meteor radars contain information about meteor source regions and patterns related to the rotation and revolution of the Earth. Using the meteor parameters from two meteor radars located in Mohe (53.5° N, 122.3° E) and Wuhan (30.6° N, 114.4° E), this study calculates the variation patterns in the meteor azimuth distribution over the two sites over 1 year. Additionally, this study introduces the variable, Max_Azi, to describe the position of the peak of azimuth distribution. The peak value of azimuth distribution is calculated by Gaussian fitting to quantify the variation patterns in azimuth distribution. This study provides complementary information on the azimuth distribution in high and middle latitudes. The results indicated that the azimuth distribution variation for the Mohe meteor radar is consistent with the Earth’s revolution model.