[1] The source regions of region-0 (R0), region-1 (R1), and region-2 (R2) field-aligned currents (FACs) were statistically determined using DMSP particle precipitation and magnetometer data. Each FAC sheet originates from more than one region in the magnetosphere, depending on the latitude and the magnetic local time (MLT). R2 originates mostly from the central plasma sheet (CPS) and boundary plasma sheet (BPS) in the morning and from the CPS, BPS, and inner magnetosphere in the afternoon, all of which are on closed field lines. Near noon, some R2 may originate from the low-latitude boundary layer (LLBL), which is located near the magnetopause and can be open or closed. R1 mostly maps to the BPS, hence on closed field lines, in morning and afternoon, but near noon, it maps mostly to the LLBL. The LLBL source region can be found more frequently in the dawn-noon sector than in the noon-dusk sector. On the other hand, R0 is located mostly on open field lines and is associated mostly with mantle precipitation. However, the mantle precipitation has a dependency on the polarity of R0. Within up-flowing R0, sometimes an upward field-aligned electric field, which accelerates electron downward and retards ion precipitation, modifies mantle distribution to look more like those of polar rain or BPS. This electric field has the opposite polarity to the background electric field that maintains charge-quasi-neutrality and that limits some solar wind electrons from entering the magnetosphere in the mantle and polar rain regions. Implications to current generation mechanisms are discussed.

The present study examines seasonal variations of large‐scale field‐aligned current (FAC) systems in terms of the dipole tilt and clock angles. Magnetic field measurements from the DMSP F7 and F12‐F15 satellites are used. This data set consists of a total of ∼185,000 auroral oval crossings, out of which ∼121,000 crossings were selected for the present analysis. Focus is placed on the latitude at the demarcation between the region 2 (R2) and region 1 (R1) currents and the intensities of these currents. It is found that the dayside FAC moves poleward and equatorward in the summer and winter hemispheres, respectively, and the nightside FAC has the opposite seasonal dependence. In the midday sector the peak‐to‐peak variation of the FAC latitude over the entire range of the dipole tilt is ∼5°, whereas it is ∼4° around midnight. In the flank sectors the average FAC latitude is higher around the solstices than around the equinoxes irrespective of hemisphere. The corresponding dependence on the dipole clock angle can actually be found for almost all local time sectors, although the peak‐to‐peak variation of the expected semiannual variation, 2° around noon and <1° in other local time sectors, is smaller than that of the annual variation except for the flank sectors. A comparison with a model magnetic field strongly suggests that the dipole tilt effect on the magnetospheric configuration is the primary cause of the annual variation, whereas the semiannual variation is inferred to reflect the fact that geomagnetic activity tends to be higher around the equinoxes. The average dayside FAC intensity is larger in the summer hemisphere than in the winter hemisphere, which can be explained in terms of the seasonal variation of the ionospheric conductivity. The dayside R1 current intensity depends more strongly on the dipole tilt than the dayside R2 current intensity, and it changes by a factor of 2–3 over the entire range of the dipole tilt angle. In contrast, the annual variation of the nightside FAC intensity is more complicated, and the nightside R2 current seems to be more intense in the winter hemisphere than in the summer hemisphere. The dependence of the FAC intensity on the dipole clock angle is less significant especially for the R1 system. Nevertheless, the result suggests that the FAC tends to be more intense around the equinoxes, which is consistent with the semiannual variation of geomagnetic activity.

Abstract.A new filtering technique for sequential data assimilation, the merging particle filter (MPF), is proposed. The MPF is devised to avoid the degeneration problem, which is inevitable in the particle filter (PF), without prohibitive computational cost. In addition, it is applicable to cases in which a nonlinear relationship exists between a state and observed data where the application of the ensemble Kalman filter (EnKF) is not effectual. In the MPF, the filtering procedure is performed based on sampling of a forecast ensemble as in the PF. However, unlike the PF, each member of a filtered ensemble is generated by merging multiple samples from the forecast ensemble such that the mean and covariance of the filtered distribution are approximately preserved. This merging of multiple samples allows the degeneration problem to be avoided. In the present study, the newly proposed MPF technique is introduced, and its performance is demonstrated experimentally.