“…For our database, we require favorable sky condition and relatively good radar coverage, especially between the 0 and 11 range gates we use in this study (covering the magnetic latitudinal range ~74°–78.5° MLAT), which are located just poleward of the most probable latitudes of the auroral poleward boundary [ Nishimura et al ., ]. The short range of these echoes indicates that the radar signals only travel a short distance before getting backscattered [see, e.g., Koustov et al ., ], so that the altitude of backscatter is expected to be in the E region ionosphere [ Gorin et al ., ; Lyons et al ., , Figure 11], and thus, the flow magnitude will be less than the full E × B drift speed and limited below the ion acoustic speed (~400 m/s) [ Haldoupis , ; Koustov et al ., ] and the mean flow angle will be shifted from the E × B direction by ~20° [ Gorin et al ., ]. However, these do not influence on our results because we only study relative changes in magnitudes of flows directed toward the auroral poleward boundary.…”