We developed a flare prediction model using machine learning, which is optimized to predict the maximum class of flares occurring in the following 24 hr. Machine learning is used to devise algorithms that can learn from and make decisions on a huge amount of data. We used solar observation data during the period 2010-2015, such as vector magnetograms, ultraviolet (UV) emission, and soft X-ray emission taken by the Solar Dynamics Observatory and the Geostationary Operational Environmental Satellite. We detected active regions (ARs) from the full-disk magnetogram, from which ∼60 features were extracted with their time differentials, including magnetic neutral lines, the current helicity, the UV brightening, and the flare history. After standardizing the feature database, we fully shuffled and randomly separated it into two for training and testing. To investigate which algorithm is best for flare prediction, we compared three machine-learning algorithms: the support vector machine, k-nearest neighbors (k-NN), and extremely randomized trees. The prediction score, the true skill statistic, was higher than 0.9 with a fully shuffled data set, which is higher than that for human forecasts. It was found that k-NN has the highest performance among the three algorithms. The ranking of the feature importance showed that previous flare activity is most effective, followed by the length of magnetic neutral lines, the unsigned magnetic flux, the area of UV brightening, and the time differentials of features over 24 hr, all of which are strongly correlated with the flux emergence dynamics in an AR.

We developed a solar flare prediction model using a deep neural network (DNN), named Deep Flare Net (DeFN). The model can calculate the probability of flares occurring in the following 24 h in each active region, which is used to determine the most likely maximum classes of flares via a binary classification (e.g., ≥M class versus

In this paper, we try to elucidate the generation mechanism of the field‐aligned current (FAC) and coexisting convection. From the comparison between the theoretical prediction and the state of numerical solution from the high‐resolution global simulation, we obtain the following conclusions about the distribution of dynamo, the magnetic field structure along the flow path that diverges Poynting flux, and energy conversion promoting the generation of electromagnetic energy. The dynamo for the region 1 FAC, which is in the high‐latitude‐side cusp‐mantle region, has a structure in which magnetic field is compressed along the convection path by the slow mode motion. The dynamo for the region 2 FAC is in the ring current region at the inner edge of the plasma sheet, and has a structure in which magnetic field is curved outward along the convection path. Under these structures, electromagnetic energy is generated from the work done by pressure gradient force, in both dynamos for the region 1 and region 2 FACs. In these generation processes of the FACs, the excitation of convection and the formation of pressure regimes occur as interdependent processes. This structure leads to a modification in the way of understanding the Dungey's convection. Generation of the FAC through the formation of pressure regimes is essential even for the case of substorm onset.

We have developed a global simulation code which gives numerical solutions having an extremely high resolution. The substorm solution obtained from this simulation code reproduces the precise features of the substorm onset in the ionosphere. It can reproduce the onset that starts from the equatorward side of the quiet arc, two step development of the onset, and the westward traveling surge (WTS) that starts 2 min after the initial brightening. Then, we investigated the counter structures in the magnetosphere that correspond to each event in the ionosphere. The structure in the magnetosphere promoting the onset is the near‐Earth dynamo in the inner magnetospheric region away from the equatorial plane. The near‐Earth dynamo is driven by the field‐aligned pressure increase due to the parallel flow associated with the squeezing, combined with equatorward field‐perpendicular flow induced by the near‐Earth neutral line (NENL). The dipolarization front is launched from the NENL associated with the convection transient from the growth phase to the expansion phase, but neither the launch nor the arrival of the dipolarization front coincides with the onset timing. The arrival of flow to the equatorial plane of the inner magnetosphere occurs 2 min after the onset, when the WTS starts to develop toward the west. The expansion phase is further developed by this flow. Looking at the present result that the onset sequence induced by the near‐Earth dynamo reproduces the details of observation quite well, we cannot avoid to conclude that the current wedge is a misleading concept.

The transition of magnetic topology is analyzed during the substorm. During the growth phase in the polar ionosphere, magnetic fields consist of three categories: closed magnetic field, open magnetic field leading to the northward interplanetary magnetic field (IMF), and open field leading to the southward IMF. While the open magnetic field region leading to the southward IMF expands consistently during the growth phase, the ionospheric onset starts inside the closed field line region. There should be triple points where three kinds of magnetic field coexist in the ionosphere. Connected to the triple points, complex magnetic topologies appear in the magnetosphere. There occur definitive changes in the global magnetic topology such as retreat of cusp nulls generated under the northward IMF, generation of plasma sheet magnetic fields with reverse curvature, generation of bifurcation areas on the magnetopause, formation of new dayside nulls, development of null lines on the flank, and generation of bifurcation areas (group of nulls) on the magnetopause. During the late growth phase, divergence Vx flow is excited in the central plasma sheet associated with the appearance of reverse curvature. The plasma sheet reconnection starts 5 min before the ionospheric onset from a localized area at X = −17.5 Re due to the reduction of the normal component Bz caused by divergence Vx flow. It occurs in the residual magnetic structure formed under the northward IMF as a three‐dimensional reconnection with a guide field By. The resulting two‐turn coil of magnetic field forms a plasmoid, which is ejected downtail.