With the increasing number of large constellations, it is crucial to accurately predict satellite positions and movements using upper atmosphere models driven by geomagnetic indices. Machine learning (ML) can quickly provide geomagnetic index predictions. However, previous research using ML to forecast Stream Interaction Region-driven (SIR-driven) geomagnetic storms primarily follows two approaches: using real-time observations at the L1 point with historical geomagnetic indices, which lacks solar source information, or using solar images as input. To forecast SIR-driven storms over longer horizons (3 days), a neural network model is constructed by incorporating coronal magnetic field physical parameters and the coronal hole features extracted from solar images (Pch ${P}_{\text{ch}}$). We first select the appropriate combination of coronal magnetic parameters for forecasting geomagnetic indices. Using the Explainable AI technique, we find that the model can learn the propagation characteristics of interactions between high- and low-speed solar wind from historical coronal parameters, which relate to the physical process of SIR formation, providing a physical basis for the approach. We then introduce Pch ${P}_{\text{ch}}$ and the historical Kp index to develop an optimal model. Based on accuracy evaluations across different activity levels and solar cycle phases, the model has proven to provide reasonable forecast results for SIR-driven events over the next 2-3 days. The model outperforms an empirical model and two ML models in terms of event detection ability for horizons longer than one day. This study offers a new approach for 3-day geomagnetic disturbance forecasting using a deep neural network.

The F10.7 index is crucial for assessing solar activity, significantly impacting communication, navigation, and satellite operations. The intrinsic complexity and variability of solar activity often result in sudden perturbations in the F10.7 index, compromising the accuracy and stability of forecasts. To address this challenge, we propose a novel prediction strategy that separately forecasts fundamental trends driven by the medium-to-long-term evolution of the solar cycle and the 27 day rotational modulation, along with transient disturbances caused by solar flares and the rapid evolution of active regions. These forecasts are then integrated to enhance overall prediction accuracy. We incorporate additional features such as the soft X-ray flare index (FI _SXR ), magnetic type of the active region (new_Mag), and X-ray background flux (XBF) to enhance the understanding of the underlying physical processes of solar activity. Our experiments, conducted using advanced forecasting models on the SG-F10.7-All data set, validate the efficacy of our proposed strategy. Notably, the iTransformer model demonstrates superior performance in both short-term and medium-term forecasting scenarios. The inclusion of FI _SXR , new_Mag, and XBF significantly improves forecasting accuracy, highlighting their importance in improving the F10.7 index predictions. Our method outperforms international models from the Space Weather Prediction Center, British Geological Survey, and Collecte Localisation Satellites, exhibiting greater accuracy and adaptability across various solar activity phases. This finding provides a novel approach for precise forecasting of the F10.7 index.

Accurate prediction of solar flares is essential for space weather warnings and safeguarding technological infrastructure. This study proposes a dual-stage flare prediction framework leveraging the full-disk flare index (FI). In the first stage, FI is decomposed into long-term trend and short-term disturbance components via a “trend-disturbance” decomposition strategy. The iTransformer model independently forecasts each component, which is then fused to generate high-fidelity FI predictions. The second stage develops a regression-classification architecture that maps predicted FI values to flare intensity levels, enabling comprehensive full-disk flare forecasting. Experimental results indicate that the decomposition strategy improves performance across all flare classes, reducing the mean absolute error of FI prediction to 19.066 and achieving a TSS of 0.639 and an F1 score of 0.649 for M-class flares. Tested from the solar minimum in 2019 through the 2024 solar maximum, the framework surpasses the operational forecasting capabilities of leading space weather prediction centers such as SWPC and SEPC, successfully delivering reliable 72 hr flare warnings during 2024 May events. The dual-stage framework introduces an innovative approach to improving the accuracy and reliability of flare prediction, offering a valuable reference for forecasting extreme solar activity events.

Volume visualization can not only illustrate overall distribution but also inner structure and it is an important approach for space environment research. Space environment simulation can produce several correlated variables at the same time. However, existing compressed volume rendering methods only consider reducing the redundant information in a single volume of a specific variable, not dealing with the redundant information among these variables. For space environment volume data with multi-correlated variables, based on the HVQ-1d method we propose a further improved HVQ method by compositing variable-specific levels to reduce the redundant information among these variables. The volume data associated with each variable is divided into disjoint blocks of size 4~3 initially. The blocks are represented as two levels, a mean level and a detail level. The variable-specific mean levels and detail levels are combined respectively to form a larger global mean level and a larger global detail level.To both global levels, a splitting based on a principal component analysis is applied to compute initial codebooks.Then, LBG algorithm is conducted for codebook refinement and quantization. We further take advantage of progressive rendering based on GPU for real-time interactive visualization. Our method has been tested along with HVQ and HVQ-1d on high-energy proton flux volume data, including ＞ 5, ＞ 10, ＞ 30 and ＞ 50 Me V integrated proton flux. The results of our experiments prove that the method proposed in this paper pays the least cost of quality at compression, achieves a higher decompression and rendering speed compared with HVQ and provides satisficed fidelity while ensuring interactive rendering speed.

Geostationary satellites are exposed to harsh space weather conditions, including ≥2 MeV electrons from the Earth’s radiation belts. To predict ≥2 MeV electron daily fluences at 75°W and 135°W at geostationary orbit for the following three days, long short-term memory (LSTM) network models have been developed using various parameter combinations. Based on the prediction efficiency (PE) values, the most suitable time step of inputs and best combinations of two or three input parameters of models for predictions are recommended. The highest PE values for the following three days with three input parameters were 0.801, 0.658 and 0.523 for 75°W from 1995 to August 2010, and 0.819, 0.643 and 0.508 for 135°W from 1999 to 2010. Based on yearly PE values, the performances of the above models show the solar cycle dependence. The yearly PE values are significantly inversely correlated with the sunspot number, and they vary from 0.606 to 0.859 in predicting the following day at 75°W from 1995 to 2010. We have proven that the poor yearly PE is related to relativistic electron enhancement events, and the first day of events is the most difficult to predict. Compared with previous models, our models are comparable to the top performances of previous models for the first day, and significantly improve the performance for second and third days.

太阳耀斑是一种重要的太阳爆发活动现象，表现为近乎全波段的电磁辐射增强。统计表明，太阳活动水平越高，太阳爆发越频繁，耀斑爆发的概率越大。利用1975-2007年10.7 cm流量与耀斑爆发的统计关系，建立了一种可行的全日面爆发耀斑概率的预报方法，能够实现C,M,X三种级别的耀斑在全日面爆发的概率预报。通过2008-2016年的观测数据，对模型进行了预报性能的评估，得到模型对C,M,X级耀斑发生概率的预报误差均较小，Brier评分误差分别为0.113,0.087,0.012；模型的预报性能均比平均模型有提高，对C,M,X级耀斑发生概率预报的Brier技巧评分分别为0.250,0.106,0.012。在2008-2016年未来1天耀斑预报的模型实测中，模型的预报效果与中国科学院空间环境预报中心的预报效果相当，这说明该模型在实际的空间环境预报中切实可行。

Abstract Combining the upstream solar wind observations measured by Mars Atmosphere and Volatile Evolution (MAVEN), Advanced Composition Explorer(ACE) and Deep Space Climate Observatory (DSCOVR) from October 2014 to April 2021, we investigate the statistical properties of the background solar wind at Mars and Earth. By applying an operational solar wind prediction system (Wang et al., 2018, https://doi.org/10.1051/swsc/2018025) in Space Weather Prediction Center (SEPC), we simulate the solar wind conditions and carry out a comparative analysis with observations to study our model performance. We find that our model is able to simulate the solar wind conditions upstream of Earth and Mars, corresponding to the different heliocentric distances and different levels of solar activity. Furthermore, we apply an event‐based evaluation by analyzing the high speed enhancements (HSEs), and find that the hit rate of HSEs is 70.38% and 66.37% for Earth and Mars, respectively. By predicting the HSEs at Earth (Mars), our model reaches a Mean Absolute Error (MAE) of 83.93 km/s (65.91 km/s) and 22.98 hr (21.65 hr) for maximum speed and arrival time prediction error, respectively. We also conduct a three‐month case study, from November 2020 to January 2021, analyzing solar wind conditions upstream of Earth, Mars, and measured by Tianwen‐1 (China's first Mars mission), for which our model is capable to predict the upstream solar wind conditions up to Mars.

The 3-day Kp forecast product is important and necessary for space weather forecasts. There is some essential information that can be obtained from the 3-day Kp forecast product, such as the start time of the geomagnetic storm, the maximum storm level, and the storm duration. In this study, we aimed to predict the next 3-day Kp index based on the previous Kp time series and SDO/AIA 193 Å images. We prepared datasets from May 2010 to December 2019 for training and datasets from January 2020 to October 2022 for testing. The similarity parameters of the previous and current geomagnetic conditions between the samples are calculated and analyzed. We assumed that the paired samples with high-similarity parameters of the previous and current geomagnetic conditions would also have high-similarity parameters of the next 3-day geomagnetic conditions. Based on the assumption, we selected the three best similarity parameters through the feature selection process and adopted the scalable tree boosting system (XGBoost) to develop a prediction model. It took the similarity parameters of the previous and current geomagnetic conditions as input and provided the best match sample from the training subset as a forecast. For the next 3-day non-storm (maximum Kp < 5) prediction period, our model reached an F1-score of 0.96. For the next 3-day storm (maximum Kp ≥ 5) prediction period, our model reached an F1-score of 0.82, a recall of 0.70, and a precision of 0.98.

针对高层大气密度预报和轨道预报业务中对新型太阳紫外辐射指数E10.7的需求，基于TIMED-SEE观测仪器提供的0.1～105 nm太阳辐射强度数据，开展了E10.7指数反演和中期预报研究。 E10.7指数是太阳光谱中波长为0.1～105 nm的辐射流量，单位与F10.7指数相同（sfu，1 sfu=10–22 W·m–2·Hz–1）。 TIMED-SEE观测仪器提供的0.1～105 nm太阳辐射强度实测值具有高时间分辨率、延迟时间短和易获得的优势，利用最小二乘法拟合可反演出准实时的E10.7指数，均方根误差为5.445 sfu。利用高阶自回归模型对E10.7的中期预报效果尚佳，未来27天的预报值平均相对误差为7.83%。利用同样方法还开展了E10.7指数81天中心滑动平均值未来27天预报试验，未来27天的预报值平均相对误差仅为3.63%。

针对高层大气密度预报和轨道预报业务中对新型太阳紫外辐射指数E10.7的需求,基于TIMED-SEE观测仪器提供的0.1～105 nm太阳辐射强度数据,开展了E10.7指数反演和中期预报研究.E10.7指数是太阳光谱中波长为0.1～105 nm的辐射流量,单位与F10.7指数相同(sfu,1 sfu=10–22 W·m–2·Hz–1).TIMED-SEE观测仪器提供的0.1～105 nm太阳辐射强度实测值具有高时间分辨率、延迟时间短和易获得的优势,利用最小二乘法拟合可反演出准实时的E10.7指数,均方根误差为5.445 sfu.利用高阶自回归模型对E10.7的中期预报效果尚佳,未来27天的预报值平均相对误差为7.83％.利用同样方法还开展了E10.7指数81天中心滑动平均值未来27天预报试验,未来27天的预报值平均相对误差仅为3.63％.

The solar wind is the direct cause of the geomagnetic disturbance. In this paper, based on the feature selection and similarity algorithm of machine learning, a recommended model is established to search for cases whose characteristics are similar to the current solar wind in historical solar wind data, and to obtain the prediction of the geomagnetic Kp index. Tested on 120 solar wind cases randomly selected from 1998 to 2019, the results show that the solar wind cases which have similar geomagnetic effects to the input solar wind can be worked out successfully by proposed model . And the root mean square error between the Kp index of the optimal case recommended by the model and the actual value is 0.79, and the correlation coefficient is 0.93. Different from traditional forecast models, the proposed recommended model in this paper can not only provide a geomagnetic Kp index as a forecast, but also give a clearer and more intuitive comparison of the changes between the solar wind characteristic parameters according to the time series. Even because the historical events have already happened, we can artificially find more dimensional information of the similar historical cases, which makes forecasters better combine their own experience inKp index forecasting.

We propose a compression and rendering method for time-varying space environment volumes, by incorporating compression in temporal dimension into HVQ, based on the temporal coherence and cyclical change characteristics of space environment. We incorporate partitions and weights into vector quantization to give more consideration to concerned period data. We take advantage of the codebook-retraining algorithm to speed up the quantization of sequences with similar space environment condition. Furthermore, we employ the progressive rendering based on GPU for real-time interactive visualization. The results of our experiments prove that the method proposed in this paper can improve the compression rate, reduce the cost of reconstruction quality for compression significantly, and also increase rendering speed in space environment domain.

In predicting the collision of space debris, the propagated orbital uncertainty may not follow a Gaussian distribution if the initial orbital uncertainty is large or the propagation time is long. In this paper, a Gaussian mixture uncertainty propagation method developed by (Psiaki et al., 2015) is used to calculate the collision probability. The initial Gaussian distribution is fitted by the weighted Gaussian mixture components. The linear matrix inequality is optimized to prevent the covariance matrix of Gaussian mixture components from being too small, and an appropriate number of Gaussian mixture components is used to approximate the initial orbital covariance. At the same time, this paper provides a method to calculate the collision probability of two objects in which a Gaussian mixture is used to represent the distribution of orbital uncertainty. The linear method and the unscented Kalman filter (UKF) method for propagating the Gaussian covariance are analysed. The results of numerical simulations show that compared with the linear covariance propagation method, UKF method, and high-precision Monte Carlo covariance propagation method for space objects with a large initial orbital uncertainty, the Gaussian mixture method can be effectively applied to capture the non-Gaussian characteristics of the predicted non-linear orbital dynamic uncertainty. Compared with the univariate splitting method, the advantage of this Gaussian mixture method is that it does not need to search for the most nonlinear direction. The accuracy of the collision probability calculation is improved from 1.460 x 10(-3) to 1.663 x 10(-3). A comparison of the computational burden between the Gaussian mixture algorithm and Vittaldev's algo-rithm to achieve the same results is presented. The calculation burden of the Gaussian mixture method is approximately 3 times that of the univariate Gaussian method. (C) 2021 COSPAR. Published by Elsevier B.V.

Solar Active Regions (ARs) are the main source regions of solar activities. The morphology, structure, and characteristic of ARs are important factors in determining solar eruptions. Therefore, the recognition of ARs is the precondition to predict solar eruptions. SDO/HMI can continuously provide full-disk photospheric images with high temporal and spatial resolution. Referring to the method of Zhang et al.[1], we performed a study of fast automatic recognition of ARs from full-disk HMI magnetograms, which involves the intensity-based thresholding, mathematical morphological analysis and region growing. Through comparing the automatic recognition ARs with the ARs compiled by NOAA/SWPC from 2010 May to 2018 December, it is found that the number of ARs recognized automatically and the number of SWPC ARs are basically consistent in the trend of variation, and their correlation coefficient is 0.87. The total number of ARs automatically recognized is slightly less than the number of SWPC ARs. Most of the unidentified ARs are small areas, weak magnetic fields and simple magnetic structures, which are highly impossible to produce powerful eruptions to affect the space environment. This method of the automatic AR recognition from real-time HMI full-disk magnetograms can directly provide the real-time AR data for forecasting solar eruptions, and accelerate the operational application of solar eruption prediction models.

In this paper, optical target monitoring orbit determination using the Gooding algorithm based on the 400 km altitude space platform was researched. The measurement error was set to 3″ and 6″ respectively for 800 km, 1500 km altitude low Earth orbits and 36 000 km altitude geosynchronous orbit to determine the initial orbit and precise orbit. Simulation results show that the initial orbit determination accuracy of the 4~15 min arc is about 10 km, and the 1~2 min arc is about 100 km when the measurement data error is between 3″ and 6″. The error of 15 min initial determination arc is 100 m. As the arc is less than 10 min, the improved error accuracy of the orbit is in the order of km scale. As the measurement data error is 3″, the initial orbit determination accuracy of the 15~20 min arc is about tens of kilometers, and that of the 8~10 min arc is 100 km. The improved orbit has an error of km. When the measurement data error is 6″, the accuracy of the initial orbit determination of the 20 min arc is at the magnitude of 10 km, the accuracy of the initial orbit determination of the 8~15 min arc is at the magnitude of 100 km, and the error accuracy of the improved orbit is at the magnitude of 10 km.

The energetic electrons in the Earth’s radiation belt, known as “killer electrons”, are one of the crucial factors for the safety of geostationary satellites. Geostationary satellites at different longitudes encounter different energetic electron environments. However, organizations of space weather prediction usually only display the real-time ≥2 MeV electron fluxes and the predictions of ≥2 MeV electron fluxes or daily fluences within the next 1–3 days by models at one location in GEO orbit. In this study, the relationship of ≥2 MeV electron fluxes at different longitudes is investigated based on observations from GOES satellites, and the relevant models are developed. Based on the observations from GOES-10 and GOES-12 after calibration verification, the ratios of the ≥2 MeV electron daily fluences at 135° W to those at 75° W are mainly in the range from 1.0 to 4.0, with an average of 1.92. The models with various combinations of two or three input parameters are developed by the fully connected neural network for the relationship between ≥2 MeV electron fluxes at 135° W and 75° W in GEO orbit. According to the prediction efficiency (PE), the model only using log10 (fluxes) and MLT from GOES-10 (135° W), whose PE can reach 0.920, has the best performance to predict ≥2 MeV electron fluxes at the locations of GOES-12 (75° W). Its PE is larger than that (0.882) of the linear model using log10 (fluxes four hours ahead) from GOES-10 (135° W). We also develop models for the relationship between ≥2 MeV electron fluxes at 75° W and at variable longitudes between 95.8° W and 114.9° W in GEO orbit by the fully connected neural network. The PE values of these models are larger than 0.90. These models realize the predictions of ≥2 MeV electron fluxes at arbitrary longitude between 95.8° W and 114.9° W in GEO orbit.

Visualization has been widely applied in space environment domain. However, compressed volume rendering algorithms based on VQ are concerned on fidelity and compression rate, not combined with specific application. To fulfill the specific visualization requirements for space environment volume data, an application-driven compression and rendering algorithm is proposed, which is Weight Based Hierarchical Vector Quantization (WHVQ). The volume data is initially partitioned into disjoint 43 blocks. Weights are assigned to the blocks according to their importance. The blocks are then decomposed into a three level hierarchical representation and each block is represented by a mean value and two detail vectors. To the top two levels, a splitting based on principal component analysis and weight is adopted to form their initial codebooks. Then, LBG algorithm based on weight is conducted for codebook refinement and quantization. The experimental results show that WHVQ is able to improve the quality of reconstruction in interested area on the premise of the good overall fidelity.

Based on the recalibrated data of >16-MeV proton omnidirectional integral fluxes obtained from NOAA POES and EUMETSAT MetOp satellites during the period from 1978 to 2014, solar cycle phase lags, the variations of proton radiation belt, and the relationships between trapped proton fluxes and F10.7 flux, cosmic ray, or sunspot number are investigated in detail. It is found that (a) the solar cycle phase lags between energetic proton fluxes and sunspot number are within 2.5 years, dependent on solar cycle, L-m, B/B-0, and proton energy; (b) at the magnetic equator, the ratios of the maximal to minimal values of 4-month smoothed monthly >16-MeV proton fluxes during the period from 1978 to 2014 increase and then decrease with increasing L-m, showing the maximal value similar to 11.24 at L-m = 1.15; and (c) at a magnetic field line with L-m <= 1.15, the ratios of the maximal values in 1987 to the minimal values in 2002 for yearly >16-MeV proton fluxes decrease with increasing B/B-0, while at a magnetic field line with L-m >= 1.16, they increase and then decrease with increasing B/B-0, and the maximal ratio for all magnetic field lines can reach above 20. An exponential equation with an offset is feasible to describe the relationship between the logarithm of >16-MeV proton fluxes and F10.7 flux. It can be readily used to improve the NOAAPRO model.

Based on AE8 electron radiation belt model and various geomagnetic field models, we analyzed the effects of geomagnetic field models, solar wind, geomagnetic disturbance and the pointing of geomagnetic axis on >= 2 MeV electron distribution at the geostationary orbit, examined the differences of >= 2 MeV electron distribution at different longitudes of the geostationary orbit, and made a comparison between the results from models and the observations from GOES satellites. It is shown that the results of >= 2 MeV electron fluxes at the geostationary orbit from AE8+IGRF+T96 model are better than those from AE8+IGRF+OPQ77 model or AE8 +IGRF +T89 model, most of qualitative results from AE8 + IGRF + T96 model are consistent with observations from GOES satellites, and >= 2 MeV electron fluxes at the geostationary orbit from AE8+IGRF+T96 model have a good negative correlation with magnetic shell parameter Lm and a good positive correlation with local magnetic field B. Based on the AE8+IGRF+T96 model, >= 2 MeV electron fluxes at the geostationary orbit each minute during 2010 are calculated by the assumption of constant solar wind conditions and geomagnetic disturbances. The data reveal that the ratios of the maximal to minimal values of >= 2 MeV electron fluxes at the geostationary orbit each minute vary from 2. 50 to 7. 51 in one year, with a main period of 1 day and the variation of ratios each day more than 3, the ratios of the maximal to minimal values of >= 2 MeV electron fluxes for any geostationary satellite each day are within the range from 2. 98 to 6. 00 in one year, varying with the time and the location, and the maximum and minimum values of >= 2 MeV electron daily fluences for the geostationary satellites at the same day in one year appear near 170 degrees W and 70 degrees W, respectively, with their the ratios varying from 1. 86-2. 13. The variations of >= 2 MeV electron distribution at the geostationary orbit above are mainly due to Lm, and the effect of B/B-o is less than 5%, where B-o is the minimal value of magnetic field at any field line. Therefore, the influence of geomagnetic field structure, especially the Lm parameter, should be considered for developing the model of >= 2 MeV high energy electron flux distribution at the geostationary orbit.

Collision warning and avoidance is the main method of the satellite risk mitigation to the catalog debris. The warning precision and confidence are the main problems of the current collision warning work. This paper first put forward the problems on the collision warning work which include the data, model and error cope methods, and then introduced the collision warning work flow and development in NSSC (National Space Science Center, Chinese Academy of Sciences). Some refined work were introduced from two aspects: the first is to improve orbit prediction precision, such as the TLE precision improvement, drag coefficient analysis, the space environment effect to the prediction accuracy; The other is to refine the collision probability computation, such as the covariance analysis. These studies could improve the warning confidence to some extent.