Prediction and variation of the auroral oval boundary based on a deep learning model and space physical parameters

Han, Yiyuan; Han, Bing; Hu, Zejun; Gao, Xinbo; Zhang, Lixia; Yang, Huigen; Li, Bin

doi:10.5194/npg-27-11-2020

Cited by 8 publications

(5 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on photographs of the aurora, Feldstein and Starkov (1967) demonstrated how the average oval moves to lower latitudes and becomes wider as the level of geomagnetic activity increases. This average response to the level of activity has since been confirmed in multiple studies (e.g., Carbary et al, 2003;Han et al, 2020;Hu et al, 2017;Milan et al, 2010;Starkov, 1994;Y. Zhang & Paxton, 2008) and it has been reported that the radius of the oval is larger when the ring current is strong (Milan, 2009;Milan, Hutchinson, et al, 2009).…”

Section: Introductionsupporting

confidence: 65%

“…Several studies have made latitudinal intensity profiles of the auroral intensity through binning or curve fitting (typically Gaussian or double Gaussian) and determined boundary locations using global threshold values (e.g., Brittnacher et al., 1999; Frank & Craven, 1988; Mende et al., 2003; Ohma et al., 2018), a fixed fraction of the maximum intensity (e.g., Kauristie et al., 1999), variable thresholds based on noise levels (e.g., Gjerloev et al., 2008) or by using the width of the fitted functions (e.g., Boakes et al., 2008; Carbary et al., 2003; Chisham et al., 2022; Longden et al., 2010). More recently, image segmentation techniques based on clustering methods or deep learning have been applied to identify auroral boundaries (Ding et al., 2017; Han et al., 2020; Hu et al., 2017; Tian et al., 2020).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Robust Estimates of Spatiotemporal Variations in the Auroral Boundaries Derived From Global UV Imaging

Ohma,

Laundal,

Madelaire

et al. 2024

JGR Space Physics

View full text Add to dashboard Cite

The aurora often appears as an approximately oval shape surrounding the magnetic poles, and is a visible manifestation of the intricate coupling between the Earth's upper atmosphere and the near‐Earth space environment. While the average size of the auroral oval increases with geomagnetic activity, the instantaneous shape and size of the aurora is highly dynamic. The identification of auroral boundaries holds significant value in space physics, as it serves to define and differentiate regions within the magnetosphere connected to the aurora by magnetic field lines. In this work, we demonstrate a new method to estimate the spatiotemporal variations of the poleward and equatorward boundaries in global UV images. We apply our method, which is robust against outliers and occasional bad data, to 2.5 years of UV imagery from the Imager for Magnetopause‐to‐Aurora Global Exploration satellite. The resulting data set is compared to recently published boundaries based on the same images (Chisham et al., 2022, https://doi.org/10.1029/2022JA030622), and shown to give consistent results on average. Our data set reveals a root mean square boundary normal velocity of 149 m/s for the poleward boundary and 96 m/s for the equatorward boundary and the velocities are shown to be stronger on the nightside than on the dayside. Interestingly, our findings demonstrate an absence of correlation between the amount of open magnetic flux and the amount of flux enclosed within the auroral oval.

show abstract

Section: Introductionsupporting

confidence: 65%

Section: Introductionmentioning

confidence: 99%

Robust Estimates of Spatiotemporal Variations in the Auroral Boundaries Derived From Global UV Imaging

Ohma,

Laundal,

Madelaire

et al. 2024

JGR Space Physics

View full text Add to dashboard Cite

show abstract

“…Because different parameters represent different magnetospheric dynamics processes, more input parameters mean that the model has more physical processes to drive at the same time, which is closer to the actual situation. In addition, we also find that the three components of IMF have similar effects on the performance However, in previous study (Han et al, 2020), we constructed an auroral boundaries model with multi input parameters based on the deep learning network, which is constructed by a two-layer restricted Boltzmann machine network (Hinton et al, 2006;Yu & Deng, 2011) and a radial basis function network (Łukaszyk, 2004). Eighteen space environment parameters are used for the input parameters of the model according to different combinations, and the effectiveness of the model with different input parameters is tested respectively.…”

Section: Discussionmentioning

confidence: 93%

“…However, in previous study (Han et al., 2020), we constructed an auroral boundaries model with multi input parameters based on the deep learning network, which is constructed by a two‐layer restricted Boltzmann machine network (Hinton et al., 2006; Yu & Deng, 2011) and a radial basis function network (Łukaszyk, 2004). Eighteen space environment parameters are used for the input parameters of the model according to different combinations, and the effectiveness of the model with different input parameters is tested respectively.…”

Section: Discussionmentioning

confidence: 99%

Modeling of Ultraviolet Aurora Intensity Associated With Interplanetary and Geomagnetic Parameters Based on Neural Networks

Han

Zhang³

et al. 2021

Space Weather

Self Cite

View full text Add to dashboard Cite

Solar wind high-energy particles enter earth's north and south poles regions along the geomagnetic field lines, colliding with atoms and molecules in the high-altitude atmosphere to excite aurora. These auroras mainly occur in the oval ring region centered on the Earth's magnetic poles, which is also named as "auroral oval" (Akasofu, 1964;Feldstein, 1963). The intensity and spatial position of the aurora ovals show the distribution of particle precipitation, which are important physical quantities in magnetosphere dynamics and space environment and have a dramatic influence on the magnetosphere, ionosphere and thermosphere. For example, particle precipitation will not only cause additional ionization in the high latitude ionosphere to absorb and disturb the radio propagation but also increase Joule heating to change the convection and composition of the atmosphere. The changes in atmospheric convection and composition will further cause ionospheric disturbances (Zhang & Paxton, 2008). Therefore, it is of great significance to establish an accurate auroral intensity model to understanding the particle energy input at high latitudes the simulation of

show abstract

“…To automatic detection, auroral oval scientists have made significant efforts. Han et al [46] suggested machine learning for Ultraviolet Imager data. Vasiliev et al [47] involved computer vision to estimate oval locations from ROTI maps.…”

Section: Discussionmentioning

confidence: 99%

Auroral Oval Boundary Dynamics on the Nature of Geomagnetic Storm

Edemskiy

Yasyukevich

2022

Remote Sensing

View full text Add to dashboard Cite

During emergency events, we could significantly depend on the stable operation of radio communication, navigation, and radars. The ionosphere, especially its auroral regions, significantly influences radio systems, which is why scientists and engineers create systems to monitor these regions. Using data from the global GNSS network, we analyzed the 10 strongest magnetic storms of solar cycle 24: five coronal mass ejection-driven (CME-driven) and five high-speed stream-driven (HSS-driven) storms. The analysis was based on the calculation of the standard deviation of the total electron content (TEC) derivative (rate of TEC index, ROTI). Under all the storms, the ROTI featured similar dynamics: the average ROTI reaches the highest values during the main phase, and the higher the intensity is, the more intense and equatorward the average ROTI registered. The highest cross-correlations are observed with a lag of 1 h, between the IMF z-component Bz and the magnetic latitude where the highest ROTI values appear. The auroral electrojet (SME index) shows the highest impact on the ROTI dynamics. An increase in the space weather indices (in absolute value) is accompanied by a decrease in the latitude where the maximal ROTI occurs. We found that the peculiarities of a storm affect the ROTI dynamics: all the CME-driven storms feature a high cross-correlation (>0.75) between the IMF z-component Bz and the magnetic latitude where the highest ROTI appears, while the HSS-driven storms feature a lower cross-correlation (<0.75) between them. The difference in duration of similar (by maximal values of geomagnetic indices) HSS- and CME-driven storms could produce differences in the highest ROTI values. Correlations show that compared to HSS-driven storms, CME-driven ones more directly impact the ROTI values and locations of regions with a high ROTI.

show abstract

Prediction and variation of the auroral oval boundary based on a deep learning model and space physical parameters

Cited by 8 publications

References 31 publications

Robust Estimates of Spatiotemporal Variations in the Auroral Boundaries Derived From Global UV Imaging

Robust Estimates of Spatiotemporal Variations in the Auroral Boundaries Derived From Global UV Imaging

Modeling of Ultraviolet Aurora Intensity Associated With Interplanetary and Geomagnetic Parameters Based on Neural Networks

Auroral Oval Boundary Dynamics on the Nature of Geomagnetic Storm

Contact Info

Product

Resources

About