5 March 2024 in Astronomy & Astrophysics
Context.The classification of galaxy morphology is among the most active fields in astronomical research today. With the development of artificial intelligence technology, deep learning is a useful tool in the classification of the morphology of galaxies and significant progress has been made in this domain. However, there is still some room for improvement in terms of classification accuracy, automation, and related issues.Aims.Convolutional vision Transformer (CvT) is an improved version of the Vision Transformer (ViT) model. It improves the performance of the ViT model by introducing a convolutional neural network (CNN). This study explores the performance of the CvT model in the area of galaxy morphology classification.Methods.In this work, the CvT model was applied, for the first time, in a five-class classification task of galaxy morphology. We added different types and degrees of noise to the original galaxy images to verify that the CvT model achieves good classification performance, even in galaxy images with low signal-to-noise ratios (S/Ns). Then, we also validated the classification performance of the CvT model for galaxy images at different redshifts based on the low-redshift dataset GZ2 and the high-redshift dataset Galaxy Zoo CANDELS. In addition, we visualized and analyzed the classification results of the CvT model based on the t-distributed stochastic neighborhood -embedding (t-SNE) algorithm.Results.We find that (1) compared with other five-class classification models of galaxy morphology based on CNN models, the average accuracy, precision, recall, and F1_score evaluation metrics of the CvT classification model are all higher than 98%, which is an improvement of at least 1% compared with those based on CNNs; (2) the classification visualization results show that different categories of galaxies are separated from each other in multi-dimensional space.Conclusions.The application of the CvT model to the classification study of galaxy morphology is a novel undertaking that carries important implications for future studies.
1 July 2024 in The Astrophysical Journal
Forecasting the amplitude and timing of the sunspot cycle is highly important for solar physics and space weather applications, but high-precision prediction of solar magnetic activity has remained an outstanding challenge. The Informer model, as the most advanced deep learning technique, is an ideal approach for predicting solar activity cycle. Using the whole-disk sunspot numbers (SSNs) between 1749 and 2023 and the hemispheric SSNs between 1992 and 2023, the amplitudes and timings of Solar Cycles 25 and 26 are predicted by the Informer model. The main results are the following: (1) the activity levels of Solar Cycles 25 and 26 continue being weak-moderate cycles with their strengths stronger than Solar Cycle 24, implying that the long-term solar variability is significantly modulated in length and magnitude by the Gleissberg century cycle; (2) the Gnevyshev peaks of Solar Cycles 25 and 26 are clearly observed with a higher value in the second peak, suggesting that the numbers of the large sunspot groups are greater compared to the small sunspot groups in these two cycles; and (3) during Solar Cycle 25, the activity level in the southern hemisphere is predicted to be stronger than that in the northern one, revealing significant asymmetry and asynchronization between the two hemispheres. Our analysis results show that solar cycle predictions can be made more accurate if performed separately for each hemisphere. Furthermore, Solar Cycles 25 and 26 are likely to be weak-moderate cycles, in agreement with the precursor-based and model-based prediction methods.
24 February 2022 in The Astrophysical Journal Supplement Series
The identification and analysis of different variable sources is a hot topic in astrophysical research. The Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) spectroscopic survey has accumulated a mass of spectral data but contains no information about variable sources. Although a few related studies present variable source catalogs for the LAMOST, the studies still have a few deficiencies regarding the type and number of variable sources identified. In this study, we present a statistical modeling approach to identify variable source candidates. We first cross-match the Kepler, Sloan Digital Sky Survey, and Zwicky Transient Facility catalogs to obtain light-curve data of variable and nonvariable sources. The data are then modeled statistically using commonly used variability parameters. Then, an optimal variable source identification model is determined using the Receiver Operating Characteristic curve and four credible evaluation indices such as precision, accuracy, recall, and F1-score. Based on this identification model, a catalog of LAMOST variable sources (including 631,769 variable source candidates with a probability greater than 95%, and so on) is obtained. To validate the correctness of the catalog, we perform a two-by-two cross-comparison with the Gaia catalog and other published variable source catalogs. We achieve the correct rate ranging from 50% to 100%. Among the 123,756 sources cross-matched, our variable source catalog identifies 85,669 with a correct rate of 69%, which indicates that the variable source catalog presented in this study is credible.
19 February 2025 in The Astrophysical Journal Supplement Series
Coronal mass ejections (CMEs), the large-scale eruptions of plasma and magnetic fields from the solar corona, play a critical role in space weather and solar–terrestrial interactions. Understanding the periodicities of CMEs is essential for predicting their occurrences and impacts, but their quasiperiodic variations in different cycles, at different latitudes, and for different angular widths, are still unclear. In the present work, we classify CMEs based on their angular widths and latitudinal distribution, using data from the version 2.0 of the Coordinated Data Analysis Workshop catalog (1996 January–2024 March). Employing Fourier and wavelet transforms, we systematically analyze the quasiperiodic variations in the occurrence rates of CMEs. The main results show the following: (1) A lesser number of periodicities is found in high-latitude CMEs while the low-latitude CMEs show a greater number of periodicities; (2) for the periodic variations, significant differences exist between different types of CMEs across latitudes. Narrow and normal CMEs exhibit similar periodicities, whereas halo CMEs show a different periodic pattern. (3) Additional periodicities were identified for all CMEs, which occur exclusively at low latitudes or during specific cycles, potentially as a result of the noticeable heliospheric dynamic pressure after the maximum time of cycle 23. (4) The identified dominant periodicities may be associated with magneto-Rossby waves at the tachocline. This mechanism can effectively explain the periodic behavior of the observed CMEs. Our analysis results might be useful for better understanding the long-term variations of CMEs during different cycles and at different latitudes, and could also have implications for the geoeffectiveness predictions of CMEs.
1 January 2025 in Astronomical Techniques and Instruments
Solar activity plays an important role in influencing space weather, making it important to understand numerous aspects of spatial and temporal variations in the Sun's radiative output. High-performance deep learning models and long-term observational records of sunspot relative numbers are essential for solar cycle forecasting. Using the multivariate time series of monthly sunspot relative numbers provided by the National Astronomical Observatory of Japan and two Informer-based models, we forecast the amplitude and timing of solar cycles 25 and 26. The main results are as follows: (1) The maximum amplitude of solar cycle 25 is higher than the previous solar cycle 24 and the following solar cycle 26, suggesting that the long-term oscillatory variation of sunspot magnetic fields is related to the roughly centennial Gleissberg cyclicity. (2) Solar cycles 25 and 26 exhibit a pronounced Gnevyshev gap, which might be caused by two non-coincident peaks resulting from solar magnetic flux transported by meridional circulation and mid-latitude diffusion in the convection zone. (3) Hemispheric prediction of sunspot activity reveals a significant northsouth asynchrony, with activity level of the Sun being more intense in the southern hemisphere. These results are consistent with expectations derived from precursor methods and dynamo theories, and further provide evidence for internal changes in solar magnetic field during the decay of the Modern Maximum.
17 April 2025 in Space Weather
Solar flares, the significant indicators of solar activity, have an impact on Earth's satellites and communication systems. Accurate prediction of solar flare events is crucial for mitigating these effects. In this work, we use multiple data sources, including Geostationary Operational Environmental Satellites soft X‐ray flare flux and the solar flare index, to forecast solar flare activity during Solar Cycle 25 (SC25). Our results show that: (a) The north‐south asymmetry of solar flare activity during SC25 is well revealed, and the southern hemisphere is greater than the northern one. (b) The Gnevyshev peaks of chromospheric flare activity are clearly identified and they are deeper than other atmospheric activity indicators. The different timescales of solar flare responses to the geomagnetic and interplanetary magnetic fields may be the cause of the Gnevyshev peaks. (c) Chromospheric flare activity lags behind photospheric sunspot activity, indicating that changes in sunspot activity precede flare events. (d) The activity level of SC25 is influenced by the modulating effect of the Gleissberg Cycle, as supported by geomagnetic precursor indices. These results offer valuable insights into the temporal and spatial distribution of solar flare activity during SC25.