Using electric field measurements as inputs, the assimilative physics-based ionospheric model (PBMOD) successfully reproduced density depletions observed at early morning local times during four consecutive orbits of the Communication/Navigation Outage Forecasting System (C/NOFS) satellite on 17 June 2008. However, the PBMOD running with plasma drift data from empirical models as inputs predicted neither plasma depletions nor irregularities on this day. Coincident over flights of a large depletion by C/NOFS and the DMSP-F17 satellite allow estimates of its longitudinal and latitudinal scale sizes. The satellite-based estimates are shown to be in reasonable agreement with PBMOD predictions. The model's reproduction of observed temporal and spatial distributions of plasma depletions suggests that our assimilative technique can be used to enhance space-weather forecasts.
 Properties of low-latitude plasma irregularities, commonly observed at post-sunset local times, have been studied extensively using ground and satellite-based measurements [e.g., Fejer and Kelley, 1980; Fejer, 1997]. The pre-reversal enhancement of upward plasma drifts, due to eastward polarization electric fields, is the primary cause of these irregularities [e.g., Richmond, 1995]. On the other hand, reported plasma depletions at post-midnight and early morning local times are scarce. The majority of reported events occurred during storm time conditions [e.g., Burke, 1979; Su et al., 2004]. In a rare occurrence during solar maximum, ROCSAT-1 simultaneously observed an intense reduction in ion density with super-cool temperatures, accompanied by large downward, eastward, and field-aligned flows near the dawn meridian [Su et al., 2004]. Burke  suggested that dawn bubbles may be caused by substorm-generated eastward electric fields at the equator or by travelling ionospheric disturbances. In a theoretical/numerical study, Eccles  showed a persistent 10 m s−1 pre-sunrise uplift of the ionosphere due to gravity-driven current. In a statistical study by Sobral et al. , two cases were found between 1980 and 1992 in the post-midnight sector during quiet times in which plasma depletions were observed. C/NOFS frequently encounters deep plasma depletions at the dawn terminator, particularly in June and September 2008. We have analyzed density and electric field irregularities at post-midnight local times during a 10-day period in June 2008 during which a fast stream in the solar wind passed the Advanced Composition Explorer (ACE) satellite. This paper describes equatorial ionospheric responses observed in the early morning local times on 17 June 2008 while a solar wind fast stream was passing Earth for comparison with predictions of the physics-based ionospheric model (PBMOD) [Retterer, 2005; Retterer et al., 2005]. PBMOD operates in both assimilative and climatological modes. The former mode is done with electric field measurements from the Vector Electric Field Instrument (VEFI) on C/NOFS. The latter has no electric-field information beyond that provided by climatological models [e.g., Scherliess and Fejer, 1999]. Plasma densities obtained from both types of simulation are compared with measurements by the Planar Langmuir Probe (PLP) on C/NOFS.
2. Model Description
 The simulation tool used in this study was developed by Retterer  over the last decade at the Air Force Research Laboratory to support the C/NOFS mission analysis. This 4-D model of the low-latitude ionosphere solves the continuity equation as functions of position and time along magnetic field lines over a range of apex altitudes. The climatological version of the simulation is carried out with initial conditions such as neutral winds and temperatures, ion production and loss rates, and plasma density and temperature prescribed by various empirical models [Retterer et al., 2005]. The assimilative version input data consist of available satellite and/or ground measurements as inputs. In this study, the only ingested data are E × B drift velocities calculated from VEFI electric field measurements and the International Geomagnetic Reference Field (IGRF) model [Maus et al., 2005]. We note that the VEFI instrument also provides magnetic field measurements. The magnetic field strength obtained from IGRF is within 1% of the measured value. Whenever electric field data were unavailable, such as simulation regions far away from the spacecraft trajectory, we used climatological values from an empirical drift model based on Jicamarca radar and the Atmospheric Explorer satellite data [Scherliess and Fejer, 1999]. Retterer  provides a fuller description of the assimilative procedures.
 Although PBMOD was designed as a tool to study both coarse-scale global ionospheric structures and fine-scale localized plasma turbulences, this paper focuses only on large-scale density depletions observed at post-midnight local times on 17 June 2008 (day of year 169). While the average Kp index was ∼2 on this solar-minimum day, the auroral electroject (AE) index showed large amplitude variations typical of fast stream intervals [Tsurutani and Gonzalez, 1987]. The simulated low-latitude ionosphere is divided into 36 longitudinal sectors of 10° width. The latitudinal grid has 201 steps along each field line. The vertical grid has 60 points varying in size and ranges for field-line apex altitudes between 140 and 4000 km. The simulation time step is 6 min. Throughout this report simulation results and satellite observations are presented in geographic coordinates.
 Although PBMOD was run in both the assimilative and climatological modes for the entire day of 17 June 2008, here we focus on four one-hour segments that include consecutive C/NOFS nightside passes labeled 912–915 in Figures 1a–1d. Black, blue, and red lines in Figures 1a–1d represent C/NOFS observations, climatology results, and assimilative model outputs, respectively. Ion densities are plotted followed by the vertical and zonal drift velocities; positive values indicate upward and eastward drifts. Before assimilating electric fields into the global model, we applied a low-pass Fourier filter [Retterer, 2005] with a 5-min window to 1-second averaged VEFI and PLP data to remove small-scale variations (represented by the black lines in top three plots of Figure 1). The vertical dashed lines in the top three plots of Figure 1 mark times when C/NOFS crossed the 06:00 LT meridian. The C/NOFS altitude and the apex altitude of the field line it occupies are represented by black and red lines in the fourth plot from the top. The satellite latitude and the magnetic equator are plotted as black and red lines in the fifth panel. Intersections of the red and black lines in plots 4 and 5 of Figure 1 mark satellite crossings of the magnetic equator. Black and red lines in Figure 1 (bottom) indicate the longitudes and local times of the spacecraft; the horizontal dotted lines denote 06:00 and 18:00 LT.
 As seen in Figure 1 (top) large-scale density depletions were observed during four consecutive satellite orbits when eastward electric fields (e.g., upward velocities) were observed by VEFI (black lines in the second plot of Figure 1) when or a few minutes before the satellite encountered the depletion. C/NOFS crossed all of the depletions presented here at early morning local times while flying at or near the magnetic equator. PBMOD reproduced the main qualitative features of the pre-dawn sector depletions (red lines) observed by PLP (black lines) based on assimilations of available electric field data obtained from VEFI (red lines in the 2nd and 3rd plots of Figure 1). Conversely, no plasma depletions or irregularities were predicted while exercising the climatological version (blue lines) with input drift velocities derived from empirical models.
 It is impossible to specify both the latitude and longitude scale sizes of plasma depletions with measurements from a single satellite. Fortunately, the F17 satellite of the Defense Meteorological Satellite Program (DMSP) crossed the same depletion structure at a higher altitude (860 km) ∼10 min prior to the C/NOFS encounter at 420 km (cf. Figure 1d). The tracks of the two satellite orbits are illustrated in Figure 2b. DMSP-F17 crossed the depletion moving north to south at 09:12–09:36 UT, while C/NOFS traversed it moving west to east at 09:24–09:48 UT. Corresponding density measurements at 1 sec resolution from DMSP and C/NOFS are plotted in Figures 2a and 2c, respectively. Note that the time scale increases downward in Figure 2a. The dotted lines in Figures 2a and 2c mark our estimates of the depletion boundaries. If we assume that the plasma density structure did not change significantly during the 10 min time offset, the dimensions of the depletion are about 14° in longitude and 21° in latitude, corresponding to the heavy-line segments of the trajectories in Figure 2b.
 A comparison of simulation results with the two satellite observations is presented in Figure 3. It shows O+ density outputs from PBMOD calculated using assimilated VEFI (Figure 3a) and climatological values for the velocity drifts (Figure 3b). Recall that each time step of PBMOD simulations is 6 min. As indicated above Figure 3, the 2D plots represent conditions at 09:36 UT and 4°S. Red stars indicate C/NOFS' location at 9:35 UT when it sampled the density minimum (295°E, 5°S); black stars mark DMSP-F17's location at 09:25UT (300°E, 5°S). Climatology-based model results, shown in Figure 3b, indicate low nightside densities in the bottom-side ionosphere. Just prior to sunrise the predicted altitude of the peak of the F2 layer (hmF2) is at ∼300 km. By contrast, assimilation-based results displayed in Figure 3a indicate that the ionosphere was lifted upward. Near times when the two satellites encountered the deep depletion hmF2 had reached ∼900 km altitude. DMSP-F17 was flying below the F2 peak. The longitudinal extent of the depletion is ∼30°. In order to demonstrate the predicted latitudinal dimension of the depletion, Figure 3c shows assimilation (red) and climatological (black) predictions of peak F-layer densities (NmF2) plotted as functions of latitude at 09:36 UT and 285°E. The climatologically predicted latitudinal separation between the two NmF2 maxima (the equatorial ionospheric anomaly) was 18°. The latitudinal width of density valley increased to 32° in assimilation-model calculations. The size of the density depletion estimated from the two satellites shown in Figure 2 is within the dimension of the depletion structure derived from the assimilative simulation results shown in Figures 3a and 3c.
 An examination of the temporal evolution of density maps derived from PBMOD (see Animation S1 of the auxiliary material) indicates that the density depletions shown in Figures 1a–1d should be regarded as passing through the same structure during four consecutive C/NOFS orbits. This structure expands or shrinks in response to the history of vertical plasma drifts (zonal electric fields). Figure 3a shows that, at longitudes > 310°E, the bottom-side of the F-layer began to refill with plasmas created after sunrise, while topside densities remained low. Several minutes later, the bottom-side density depletion (<400 km between 280°E and 310°E longitude) extended further into normally topside altitudes due to large upward velocities observed at 09:36 UT (second plot of Figure 1d). However, the longitudinal width of the depletion continued to decrease with time as its most eastward portion crossed the dawn terminator where the ionosphere was replenished with new plasmas generated by photoionization.
4. Summary and Discussion
 The assimilative version of PBMOD that had VEFI inputs reproduced qualitative large-scale features of density depletions observed near dawn by the PLP during four consecutive C/NOFS orbits on 17 June 2008. No plasma depletions or irregularities were predicted by the climatology version of the model. DMSP-F17 and C/NOFS made north-to-south and west-to-east cuts through the same density depletion structure within 10 min of each other. At the times of these encounters the latitudinal and longitudinal dimensions of the depletion are ∼21° and 14°, respectively. This observational result is within the estimated 32° by 30° size of the density depletion derived from assimilative modeling results. Consistent with observations of the depletion at 420 and 860 km, the assimilation indicated that hmF2 reached altitudes near 900 km.
 A significant advantage of the 4D physics-based modeling is to provide reasonable global information of ionospheric responses when and where in-situ satellite observations are not available. Assimilative modeling results confirm that C/NOFS sampled the same density depletion in four orbits between 04:00 and 10:00 UT. As the depletion crossed the dawn terminator it began to refill with fresh plasmas from the lower ionosphere. The depletion structure expanded and contracted in response to the history of zonal electric fields. The uplift of the ionosphere is due to an eastward electric field, while westward electric fields act to stabilize the nightside ionosphere. The cause of the large-scale dawn density depletion is similar to that of post-sunset plasma irregularities or bubbles, however, their generation mechanisms are different.
 Plasma density depletions discussed in this paper were observed during solar minimum when the average F10.7 was ∼67. The gravity-driven current suggested by Eccles  may be the dominant effect to explain the dawn depletion during extended quiet geomagnetic conditions. Upward drifts of 10–15 m s−1 at dawn due to mig × B currents, as estimated by Eccles , are considerably smaller than those inferred from VEFI measurements on 17 June 2008. We cannot rule out contributions from disturbance dynamos [Scherliess and Fejer, 1997] excited after a corotating interaction region (CIR) at the leading edge of a fast solar wind stream observed by ACE at ∼16 UT on 14 June. Large excursions of the AE index occurred at 21 UT on 16 June and at 2 UT on 17 June. Over-shielding eastward fields may be induced when Region 2 currents exceed those of Region 1 when AE decreases rapidly [Kikuchi et al., 2000]. Further investigation is needed in order to separate various contributions in the generation of eastward electric fields.
 Small-scale variations in the drift velocities have been smoothed by a low-pass filter prior to use as input for the global run of PBMOD in order to maintain a reasonable model run-time for forecasting purpose. As a result, this study is limited in scope to an examination of large density structures in the ambient ionosphere. The spatial and temporal resolution of the global PBMOD run can be enhanced with increased computing power. The small-scale density irregularities, such as equatorial bubbles or spread F, can be further investigated by the regional plume model algorithms in PBMOD [Retterer, 2005].
 The neutral wind plays an important role in the thermosphere-ionosphere dynamics. In future studies, we plan to incorporate measurements obtained from the Neutral Wind Meter on C/NOFS for PBMOD assimilation to investigate the neutral wind effect and to further validate the space weather forecasting capabilities.
 The C/NOFS mission is supported by the Air Force Research Laboratory, the Department of Defense Space Test Program, the National Aeronautics and Space Administration, the Naval Research Laboratory, and the Aerospace Corporation. The analysis was supported in part by Air Force Office of Scientific Research Task 2301SDA5, Air Force contract FA8718-08-C-0012 with Boston College, and NASA grant NNH09AK05I to the Air Force Research Laboratory.