We examine Millstone Hill incoherent scatter radar data collected over two solar cycles between 1979 and 2001 to determine average characteristics and features of storm time ion velocity and flux transport in the American sector midlatitude ionosphere. We use over 1100 radar azimuth scans identified as containing enhanced westward ion velocities associated with the subauroral polarization stream (SAPS), covering the 12–06 magnetic local time (MLT) sector and 50°–68° invariant latitude for weak to moderate disturbance levels with Dst from 50 to −200 nT. We find the magnetic latitude peak location of the SAPS flow channels decreases linearly with both Dst and MLT with a very good degree of correlation. We also examine for the first time SAPS peak westward ion fluxes, which transport material westward with magnitude between 3 × 1013 and 3 × 1014 m−2 s−1 in a manner nearly invariant to activity level. This invariance is maintained by an inverse relationship between electron density and ion velocity magnitudes with increasing Dst. Westward log ion flux and ion velocity are maximum in the dusk sector and decrease linearly with increasing MLT, smoothly varying across the dusk terminator. Finally, velocity distributions show that material in the afternoon SAPS flow is swept westward to earlier MLT values, delivering O+ flux to the cusp region. In contrast, SAPS streams in the post terminator sectors are fixed east-west in the Sun-Earth inertial frame, effectively maintaining entrained ion fluxes at the same MLT.
Foster and Burke  introduced the term subauroral polarization stream (SAPS) to describe a persistent westward ionospheric convection velocity peak equatorward and clearly separated from the auroral convection flow pattern during dusk sector geomagnetically disturbed intervals. This midlatitude ionospheric flow is broad in general (3–5 degrees wide) with typical velocity magnitudes of several hundred m s−1 or larger, associated with poleward electric field strengths >20 mV/m. Within this broader SAPS channel, narrow regions (<1°) of intense, highly structured, variable velocities up to 2–3 km/s in L shell aligned directions have been reported as subauroral ion drifts (SAID) [e.g., Anderson et al., 1991, 1993], associated with electric fields >100 mV/m [Foster and Erickson, 2000; Erickson et al., 2002].
 SAPS flow channels are located within the plasmasphere boundary layer [Carpenter and Lemaire, 2004], a region of overlap between the cool, dense plasmasphere and the hot, tenuous auroral ionosphere characterized by electron and ion precipitation and the high-latitude convection pattern. SAPS velocities are a key signature of magnetosphere/ionosphere coupling, and are magnetically associated with region 2 current closure systems as low-energy ions flow downward and close through the low-conductivity subauroral ionosphere [Rich et al., 1980; Senior and Blanc, 1984; Gkioulidou et al., 2009]. During disturbance times, these closure currents are a hallmark of the asymmetric enhanced O+ ring current in the Sun-Earth equatorial plane, driven by inner magnetosphere hot particle pressure gradients [Ridley and Liemohn, 2002; Kistler et al., 2010]. During moderate to severe geomagnetically disturbed intervals (Kp > 3), SAPS flows overlap the edge of the inner plasmasphere where storm enhanced density (SED) features reside, and deliver significant levels of ion flux westward to the cusp region [Foster et al., 2004]. This flux provides an important storm time source population for cusp ion upwelling [Andre and Yau, 1997], which has been identified as an important source of heavy, cold ions populating the radiation belts and mass loading the plasma sheet and ring current [e.g., Jordanova, 2003].
Foster and Vo  used a two solar cycle long data set from the midlatitude Millstone Hill incoherent scatter radar (42.6 N geodetic latitude, 288.5 E geodetic longitude; 54 Λ; L ∼ 3.5) to statistically characterize the strength, location, and width of the SAPS velocity feature in the North American sector as a function of activity level (Kp) and local time. The incoherent scatter radar technique [e.g., Evans, 1969] is considered the most powerful ground-based method available for remotely studying ionospheric dynamics, and produces altitude and time-dependent information on ionospheric plasma densities, velocities, temperature, and composition. Millstone Hill, with its subauroral location and wide scan radar field of view, has access to the full range of plasmasphere boundary layer phenomena and has regularly observed SAPS and SED events over North America. We report here on a study which extends the work of Foster and Vo  to examine in depth the statistical behavior of the afternoon, dusk, and evening sector SAPS peak westward ion flux as directly sampled by the Millstone Hill radar at American longitudes during disturbance periods. In the process, we derive expressions for the magnetic latitude of the SAPS peak as a function of magnetic local time (MLT) and Dst. We also examine the statistical distribution of SAPS westward velocities at different activity levels and magnetic local times.
Foster and Vo  used the Kp index, which is calculated on 3 h intervals, as a measure of magnetic activity when statistically characterizing SAPS features. In this study, we replace Kp with the hourly Dst index [Sugiura, 1964] as a magnetic activity indicator. This is appropriate for several reasons. SAPS features are driven by ring current pressure gradients in a manner described by several authors [e.g., Yeh et al., 1991; Anderson, 2004]. In particular, ring current energy flows into the ionosphere through field-aligned region 2 currents which close through the low-conductivity subauroral region, and narrow charge separation between the electron and ion precipitation boundaries leads to intense subauroral westward ion drifts. During magnetic storms, Dessler and Parker  and Sckopke  showed that a linear relationship exists between the observed depression of the geomagnetic field at low latitudes and the total ring current energy. They conclude that the Dst index is a useful measure of ring current strength, particularly in the dusk sector [e.g., Greenspan and Hamilton, 2000]. Additionally, the Dst index is calculated on an hourly cadence which is well matched to appropriate statistical data bin sizes used in this study, and this enhanced resolution (as compared to Kp) better captures the dynamics of geomagnetic storm response.
2. Subauroral SAPS Measurements Using Incoherent Scatter Radar
 The Millstone Hill incoherent scatter radar has been operated by the Massachusetts Institute of Technology for more than 4 solar cycles, and uses megawatt class UHF transmitters combined (since 1979) with a fully steerable 46 m diameter antenna. The facility's field of view is extensive and spans more than 30 degrees of latitude and 4 h of local time at F region heights, with altitude coverage between 100 and 1000 km and temporal resolution for typical ionospheric parameters of 30 to 45 minutes for a wide coverage scan. A regular yearly program of local/regional, storm alert, and internationally coordinated experiments has built up a large database of azimuth scans across the eastern North American continent in the 220° to 350° sector [e.g., Holt et al., 1983, Figure 2]. This westward direction is well positioned to observe ionospheric features along constant lines of magnetic latitude (i.e., constant L shells). Individual azimuth position measurement of plasma parameters have fine-scale temporal and spatial structure averaged out due to a typical temporal resolution of ∼30 seconds and an along-beam spatial averaging of between 100 and 150 km. Within these azimuth scans, SAPS channels appear as a high-velocity feature extending westward to early afternoon sectors with a high degree of alignment along magnetic latitude channels [Anderson, 2004].
 In this study, we use the same Millstone Hill radar database and SAPS location techniques as employed by Foster and Vo . Specifically, we use measurements from ∼9800 separate azimuth scans collected during 1979–2001 to construct a database of over 1.4 million ion velocity and plasma parameter measurements for weak to strong magnetic activity conditions, sorted by date, geographic and geomagnetic position, activity level, and local time. From this large scan database, each azimuth scan was displayed separately using velocity data from 350 to 500 km altitude. When SAPS was present, the invariant latitude of the peak westward velocity feature was recorded in the database. Each SAPS peak was identified as given by Foster and Vo  either at a location separated from the auroral westward (premidnight) or eastward (postmidnight) two-cell convection pattern, or alternately at an inflection point enhancement on the equatorward slope of the dusk sector two-cell region pattern. Since our database does not contain auroral particle measurements to aid in identifying convection boundaries, our SAPS selection criteria are by nature conservative and will miss some cases. This is particularly true when the high-latitude portion of the SAPS velocity field merges with auroral convection in late afternoon and dusk MLT sectors [Anderson et al., 1991]. Finally, outlier/bad data points were later excluded by reexamining individual SAPS events with peak velocity magnitudes and/or locations well separated from median values. The analysis arrived at a final database total of 1103 scans containing SAPS. We assume that SAPS flows are nearly L shell aligned in the region of interest as shown by Anderson , and therefore line of sight ion velocities are corrected with a magnetic direction cosine factor to yield the westward flow component. Ion composition is measured by the radar but in virtually all cases is entirely O+.
 From the large amount of SAPS information contained in the database, we concentrate in particular on the characteristics of (1) the magnetic latitude location of the SAPS peak velocity magnitude; (2) SAPS peak westward ion velocity magnitude; and (3) SAPS peak westward ion flux, which is a product of electron density and magnetically aligned peak ion velocity. A future publication will use the radar observations to examine F region Pedersen conductance in SAPS flow channels.
3. Statistical Coverage
Figure 1 plots the statistical distribution of SAPS peak velocity identifications in the Millstone Hill study database as a function of MLT (Figure 1, top) and Dst index (Figure 1, bottom). Also plotted in Figure 1 (top) is the occurrence percentage of identified SAPS events within the full Millstone Hill azimuth scan database as a function of MLT. A total of 1103 scans from the full ∼9800 azimuth scan database contain identified SAPS velocity channels, with the majority coming from Dst levels below −100 nT and occurring at times ranging from the dusk and postmidnight sectors in the 15 to 04 MLT range, peaking at 20–21 MLT. The overall SAPS radar observational database is well sampled in activity level, as it has approximately the same distribution in Dst as the Dst value probability density function from 1963–2002 given by Wanliss and Showalter .
 It should be noted that SAPS occurrence statistics in general are difficult to determine due to the nature of the radar data and analysis techniques used in this study. Specifically, the Millstone Hill radar system is not operated continually but rather on an event driven or separately coordinated campaign experiment basis. Furthermore, identification of SAPS channels in the individual scan data relies on a delineated peak velocity feature, and there may well be times when a weaker SAPS enhancement is not visible above an enhanced equatorward tail on the high-latitude convection flow. However, we have verified the conclusions of Foster and Vo  that SAPS is observed at subauroral latitudes 50 percent of the time between 20 MLT and 02 MLT for moderate to severe geomagnetic disturbance levels.
4. SAPS Characteristics
4.1. SAPS Peak Location Variation With Dst
 SAPS features are regarded as a marker for the location of the inner boundary of the ion plasma sheet and the ring current peak [Anderson et al., 1993; Goldstein, 2006; Huang and Foster, 2007]. The Millstone study database allows an investigation of the dynamics of this boundary by examining the average SAPS peak location magnetic latitude (MLAT). We find that SAPS peak location decreases uniformly as a function of both Dst and MLT with a very good degree of correlation. This confirms as in the above cited studies that as disturbance level increases, the ring current and plasma sheet ions move earthward.
Figure 2 plots the SAPS peak location MLAT versus MLT for four Dst ranges (−200 to −100 nT; N = 48, −100 to −50 nT; N = 280, −50 to 0 nT; N = 693, 0 to 50 nT; N = 45) using 1 h binning except for −200 to −100 nT, which uses 2 h bins. MLT bins containing fewer than 3 points are eliminated from consideration. Uncertainties on the peak velocity locations are calculated using the standard deviation within the bin divided by the number of bin points. For each Dst range, linear regression is applied to the MLT variation and the resulting model coefficients are shown in the color corresponding to their range. The correlation coefficient of the fit is also given as the Pearson correlation R value. This coefficient has a magnitude approaching 1 for cases where the fitted model tracks the data with a high degree of correspondence, and is relatively insensitive to any error in overall bias (e.g., error in the intercept for linear fits).
 We obtain SAPS peak location MLAT vs Dst relationships from the fitted functions as MLAT = (69.6 − 0.63 MLT) Λ for Dst of [0,50] nT, MLAT = (66.7 − 0.56 MLT) Λ for Dst of [−50,0] nT, MLAT = (65.3 − 0.69 MLT) Λ for Dst of [−100, −50] nT, and MLAT = (62.0 − 0.69 MLT) Λ for Dst of [−200, −100] nT. The slope of the linear decrease in MLAT is consistent at all activity levels at approximately −0.6 to −0.7 degrees per MLT hour. Our conclusions are in close agreement with the earlier Kp binned Millstone Hill based studies of Foster and Vo  and Huang and Foster  and the 840 km altitude DMSP solar maximum study of Wang et al. , all of whom quote SAPS peak velocity MLAT decreases between −0.6 and −0.8 MLAT degrees per MLT hour. (Although Wang et al.  find an exponential variation of SAPS peak location with MLT, the variation is statistically indistinguishable from a linear MLAT decrease over the range of Dst values sampled by the Millstone Hill SAPS database.)
4.2. SAPS Westward Peak Ion Flux
 The transport of dusk sector material westward toward the noontime cusp is a key hallmark of SAPS flow channels. For the first time, we can examine the statistical behavior of the peak ion flux directly using the Millstone study database, since at each identified SAPS peak location, the database contains directly measured electron density and line of sight ion velocity. With a velocity correction factor described in section 2, the product of these quantities is cross-field peak ion flux aligned with magnetic latitude contours. We find a remarkably consistent westward peak ion flux with a well defined MLT decrease and a magnitude which is invariant to Dst (e.g., ring current energy).
Figure 3 (top) plots a binned analysis of SAPS 350–500 km median peak westward ion flux on a log scale as a function of MLT for the same four Dst ranges and MLT time bins as used in section 4.1, using identical color coding as Figure 2 for each activity level. Bins with fewer than 3 points are eliminated from consideration, and all SAPS magnetic peak velocity locations are combined when averaging. Overall log flux values have a small variation with Dst and a well defined linear decrease as a function of MLT, with an average SAPS peak westward transport ranging from 3 × 1014 m−2 s−1 in the postnoon sector at 14 MLT to almost an order of magnitude weaker at 3 × 1013 m−2 s−1 near 04 MLT.
 By combining all Dst values, we find that the median SAPS peak westward ion flux in the American sector between [50, 68] Λ can be well described with a very high degree of correlation (R = 0.98) by the empirical equation log(ϕSAPS) = 14.6 − 0.07 MLT, where ϕSAPS is the SAPS peak ion flux, MLT is expressed in hours and flux is measured in m−2 s−1. Furthermore, the standard deviation of SAPS flux values is approximately ±0.3 in log units (i.e., factor of 2) regardless of MLT. Activity-invariant SAPS median peak westward flux and its variance as a function of MLT are plotted in Figure 4, which shows median peak westward log ion flux across all Dst values.
 We also note the invariance of SAPS peak westward ion flux variability to the passage of the solar terminator and the entry into local darkness, despite the changing nature of the E region ionosphere during this period. In the absence of precipitation, E region conductance changes from high to very low values over the course of the 13–06 MLT database span when all seasons are taken into account, as the E region density disappears postsunset due to recombination combined with loss of EUV production [Rishbeth and Garriott, 1969]. However, against a postsunset background low conductivity, fixed locations associated with SAPS could enter during some events into the upward Region 2 field aligned current region and create through precipitation a locally enhanced E region ionosphere. This implies a potentially large variability in postsunset E region conductance when considered statistically, but we conclude that these factors do not produce an effect on SAPS westward fluxes large enough to rise above the factor of 2 variability seen.
4.3. SAPS Flux Components: Electron Density and Ion Velocity
 Although SAPS peak westward ion flux is invariant to disturbance level (Figure 3), we find that SAPS peak location electron density and ion velocity have a strong Dst dependence in addition to an overall decrease with MLT from afternoon through dusk to postmidnight. Furthermore, these quantities have a significant inverse relationship, with higher-activity levels maintaining constant flux levels through larger velocities and lower-electron densities.
Figure 3 shows median SAPS peak location log electron density (Figure 3, middle) and westward ion velocity (Figure 3, bottom) using the same binning techniques as in section 4.2. (More variability is seen in velocities at 20 MLT for low activity with Dst at [0,50], but that data has relatively poor statistical coverage.) Associated uncertainties in binned electron density and westward ion velocity are approximately 15 to 35 percent and 25 to 40 percent, respectively. The highest-activity bin of [−200, −100] Dst has the largest SAPS velocities by a factor of 2.5 with an electron density decreased by almost this exact factor. Comparison (not shown) with both the International Reference Ionosphere 2007 [Bilitza and Reinisch, 2008] and Incoherent Scatter Radar Ionospheric Model [Zhang et al., 2007] ionospheric density models shows that SAPS peak location electron density values are reduced by up to a factor of 4 over their values during quiet times.
 One potential driver of the large decrease in electron density at higher velocities is a consequence of increasing ion-neutral relative velocity, leading to frictionally enhanced effective ion temperatures (through ion-neutral collisions and heat conduction) and a correspondingly enhanced O+ recombination rate with neutrals [Schunk et al., 1976; St.-Maurice and Torr, 1978]. A separate binned analysis of radar measured electron and ion temperatures (not shown) confirms an activity-dependent increase in both of these parameters of several hundred K as Dst approaches −200 nT and velocities approach 1000 m/s. Increase in ion temperatures are expected for a frictionally driven mechanism [cf. Schunk et al., 1975, Table 1]. In extreme cases, large electron temperature increases are also associated with Region 2 field aligned current footprint regions through heat flow from magnetospheric sources, which produce stable auroral red (SAR) arcs and associated phenomena [Kozyra et al., 1997; Baumgardner et al., 2008]. Although SAPS flows in general produce a level of electron temperature enhancements well short of those associated with SAR arcs, it is possible that lower level high-altitude energy sources associated with enhanced ring currents produce the electron temperature increases seen. Another possibility is the presence of anomalous wave heating which can raise electron temperature by several hundred K or more [St.-Maurice et al., 1981].
 Our electron density results are also compatible with nighttime electron density trough modeling [Schunk et al., 1976] which finds that moderate relative flow velocities ∼500 m/s and greater can produce deep reductions in density in the absence of sunlight when present for significant periods of time. Another possibility for decreasing electron density is expansion and upward transport associated with SAPS flows, particularly near the F peak during times of large velocity magnitudes [Anderson et al., 1991].
4.4. SAPS Ion Velocity Distributions
 We can gain further insight into the properties of SAPS driven flow channels by examining the distribution of peak ion velocities at different MLT values, and we find a large difference between SAPS velocity characteristics in the postnoon and evening sectors. The statistical SAPS peak velocity distribution across all Dst values in the late afternoon period between 16 and 18 MLT is plotted in Figure 5 (N = 164 points) while the velocity distribution in the evening period between 22 and 24 MLT is plotted in Figure 6 (N = 189 points). The mean, median, and standard deviation of each distribution is indicated. Also plotted as a red line is an estimate of the magnitude of the average corotation velocity over the geographic latitude span appropriate to observed peak locations.
 The results indicate that daytime SAPS streams are not only characterized by higher-peak westward velocities (mean = 730 m/s; median = 753 m/s) but also by a much broader distribution of values reaching up to 1200 m/s. In the evening period, however, the peak westward velocity distribution is more tightly clustered around lower values (mean = 468 m/s; median = 416 m/s) with a tail reaching to 1000 km/s for large disturbances (compare also Figure 3). We also note that the evening westward peak ion velocity distribution's median value is very close to the average eastward corotation velocity at SAPS latitudes.
 These results imply that, at least at the SAPS peak, evening sector streams after the day/night terminator are fixed east-west in the inertial Sun-Earth reference frame and remain nearly at the same MLT value for a 2–3 h period. In contrast, afternoon/dusk sector SAPS streams before the day/night terminator have a peak westward velocity in the majority of cases which moves material westward to earlier MLT values and ultimately toward noon. We emphasize that the results here only examine peak SAPS westward velocities. As Anderson  notes, regions under the influence of SAPS processes can also have associated poleward flows driven by zonal electric fields, which will move ionization northward even in the dusk MLT sector where flows are inertially fixed east-west (Figure 6).
 As described by Foster and Burke , SAPS is associated with magnetosphere-ionosphere interactions and ionospheric feedback in the plasmasphere boundary layer, associated with magnetospheric field aligned region 2/region 1 closure currents and the asymmetric enhanced ring current. Our study reveals a number of intriguing features of this important geospace coupling process.
5.1. Constancy of SAPS Westward Peak Ion Flux
 Our statistical studies reveal that, on average, conditions at the SAPS velocity peak efficiently and repeatedly maintain an inverse density/velocity relationship (Figure 3). At a given MLT value, the geospace system appears therefore to have an inherent regulation mechanism which results in a constant cold ionospheric peak O+ bulk flux +/− a factor of 2, independent of activity level, in a cross-field subauroral stream flowing westward toward the noontime cusp. Future modeling of the physics of ring current/ionospheric interactions should be employed to determine whether enhanced recombination within high-speed flows, upward expansion/transport, or a combination of factors (see section 4.3) is responsible for this invariant behavior.
5.2. SAPS Enhanced Westward Ionization Transport
 By integrating westward SAPS ion flux across its latitudinal width, we can estimate the net bulk transport of ionization toward the noontime cusp. Several authors have shown that SAPS velocity channels have a significant decrease in width by a factor of 2 or more as MLT increases from dusk across midnight to early morning sectors [Foster and Vo, 2002; Anderson, 2004]. The westward peak ion flux is also significantly decreasing as MLT advances in Figure 3, and the combination of these factors heavily weights the overall westward SAPS transport of subauroral ionization as occurring in the dusk sector as opposed to the midnight/postmidnight sectors. These results have been seen in prior individual storm event studies which show significant subauroral westward ion flux [Foster et al., 2004] transporting dusk sector material in a flow channel located on the poleward edge of storm enhanced density plumes [Foster et al., 2005, 2007]. However, our study confirms this finding in a repeatable, statistical sense.
5.3. MLT Dependence of SAPS Peak Velocity and Ion Flux
 There is a linear decrease in both SAPS peak westward log ion flux and ion velocity with MLT for a given disturbance level. The velocity feature has been seen in a number of studies [e.g., Foster and Vo, 2002; Wang et al., 2008] but for the first time we report the same MLT trend in westward flux. Furthermore, decreasing velocity and flux magnitudes occur even well beyond the terminator, when conductivities in the E region conjugate points of the SAPS magnetic field line have stabilized at their low nighttime values.
 Previous experimental and model results [Elkington et al., 2005; Shinbori et al., 2005; Nishimura et al., 2006, 2007] have shown that the enhanced disturbance time ring current is asymmetric with local pressure maxima in the dusk sector. These particle pressure gradients lead to a L = 2−4 radial electric field which has an amplitude that is higher in the dusk sector by more than an order of magnitude [Wygant et al., 1998; Nishimura et al., 2007]. In the presence of strong coupling region 2 closure currents, this field will map into the subauroral ionosphere as a stronger dusk poleward electric field (westward ion velocity) channel. Our peak velocity magnitude results are compatible with these findings. We emphasize that the drop in peak SAPS westward velocity should not be interpreted as the appearance of the Harang reversal in our measurements since we track the peak velocity which remains clearly equatorward of the east-west zero velocity line [cf. Foster and Vo, 2002, Figure 6].
 The behavior of the peak location SAPS westward ion flux observations are more complex as they involve both electron density and ion velocity. A detailed understanding of this observed flux morphology requires accurate modeling of both ion-neutral coupling and thermal processes in these MLT in a manner which includes realistic subauroral region 2 closure currents. These efforts are beyond the scope of this study, but should be performed in the future to correctly explain the observed morphology.
5.4. SAPS Flows Preterminator and Postterminator
 Aggregate SAPS peak westward velocity distributions are much broader in the afternoon and dusk sectors, and our study shows that many observed values are significantly larger than that required to oppose corotation. In particular, material in the stream is swept westward to earlier MLT values and delivers O+ flux to the noontime cusp region where it can participate in upwelling processes. In contrast, SAPS streams in the post terminator sectors have a median westward peak velocity magnitude almost equal to the average corotation velocity. This implies that SAPS postsunset processes, at least at their peak locations, maintain material in an east-west sense at nearly the same MLT.
 It is important to note, however, that these streams also can move poleward [Anderson, 2004] and upward [Foster et al., 2005] while their westward motion is opposing corotation so they are not likely to completely stagnate. Additionally, our study database concentrates only on the American longitude sector, where the offset between the rotational and magnetic poles remains relatively large, changing only by ∼3 degrees over the ∼30 degrees of longitude sampled by the Millstone Hill study database.
 These results have interesting and complex implications. In the absence of ion-neutral coupling processes, American sector fixed-MLT SAPS flux streams in the postterminator period will occur at the peak velocity location in the presence of a background neutral atmosphere with the same zonal parameters (e.g., tidal motions), but potentially different meridional parameters since poleward flows can occur. In general, ion-neutral coupling at the relatively high 350–500 km altitudes sampled by this study is weak, since we have seen in Figure 3 that electron density can be quite low in the dusk sector MLT range. However, even in this case, longitudinally dependent effects such as zonal tides are still likely to cause some level of background thermospheric modulations which can interact with the SAPS ionospheric motions. Furthermore, as ion velocities continue, dynamo feedback mechanisms will gradually accelerate the thermosphere and may cause wind perturbations which further affect the electric field strength.
 The relative importance of these multiple effects during SAPS events is not yet known, in particular for other longitude sectors where the offset between rotational and magnetic pole is much smaller. Detailed future studies modeling ion-neutral coupling in all local time sectors during disturbances characterized by the appearance of subauroral flow channels are strongly suggested to further understand ionospheric and thermospheric dynamics during storm time plasma redistribution in the coupled geospace system.
 The authors would like to acknowledge H. B. Vo and J. C. Foster for the initial identification of SAPS electric field measurements in Millstone Hill radar scans. J. C. Foster and L. P. Goncharenko provided useful discussions and insight. We thank one of the reviewers as well for helpful and very insightful comments. We gratefully acknowledge J. M. Holt, W. Rideout, and members of the Atmospheric Sciences Group at MIT Haystack Observatory for assembling and calibrating the long-term Millstone Hill radar database and for the Madrigal distributed database system. SAPS statistical analysis was supported under NASA Living With A Star Targeted Research and Technology Program grant NNX07AO76G. Research support for two of the authors (M. Miskin, F. Beroz) was provided by NSF Research Experiences for Undergraduates grant AST-0647787 to MIT Haystack Observatory. Radar observations and analysis at Millstone Hill are supported under Cooperative Agreement with the Massachusetts Institute of Technology by the U.S. National Science Foundation under the Geospace Facilities program within the Geosciences Directorate, Atmospheric and Geospace Sciences Division.
 Robert Lysak thanks the reviewers for their assistance in evaluating this paper.