Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 A comparative analysis of the solar wind conditions was performed for extremely disturbed event on 29–31 October 2003. It was found that the ACE and Geotail upstream monitors provided very similar data on the IMF but that plasma measurements in the SOHO CELIAS/MTOF, ACE SWEPAM, IMP 8 MIT, and Geotail CPI experiments are very different. The solar wind velocity was indirectly estimated using the time lag for propagation of such solar wind structures as interplanetary shock, Alfven waves, rotational, and tangential discontinuities from point L1 to the Earth. We found the best correspondence of the estimated velocity was with the ACE SWEPAM data, which displayed very fast (up to 2000 km/s) solar wind, while the IMP 8, Geotail, and SOHO plasma instruments are unable to measure such a fast solar wind stream. Application of the magnetopause models to a data set of numerous geosynchronous magnetopause crossings observed by GOES and LANL satellites enabled estimation of the solar wind dynamic pressure. In general the estimated pressure and density are in agreement with the solar wind plasma parameters provided by the ACE SWEPAM experiment. An estimation of the solar wind density corresponds very well to the electron density restored from the Geotail PWI data. However, during 1600–1800 UT on 29 October, 1700–1800 UT on 30 October, and 0000–0400 UT on 31 October, the estimated solar wind pressure and density are several times larger than provided by the Geotail PWI and ACE SWEPAM. A large helium abundance is considered as a possible reason for the solar wind pressure underestimation in the first case. The understated solar wind density on 30–31 October might be explained by errors in the method for restoring of the plasma data in fast solar wind (>900 km/s) accompanied with intensive fluxes (few tens of particles per cm2 s sr) of high-energy (>30 MeV) solar energetic protons.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Time interval on 29–31 October 2003 is characterized by extremely strong solar, heliospheric, and geomagnetic disturbances [Lopez et al., 2004; Veselovsky et al., 2004]. One of the greatest solar flares began at ∼1000 UT on 28 October in the center of the solar disk. The flare was associated with the launch of an extremely fast CME and intensive fluxes of relativistic nuclei. Interplanetary fluxes of >100 MeV protons increased on four orders of magnitude (see Figure 1). At ∼0600 UT on 29 October the CME foreword shock pushed the Earth's magnetosphere and great geomagnetic storm had begun. Large and short positive variation of the SYM-H index is directly associated with a strong and fast compression of the magnetosphere. The time of ∼20 hours required for the CME propagation from the Sun to the Earth permits the approximate estimation of the CME velocity of about 2000 km/s. Another CME was ejected from the large solar flare at ∼2040 UT on 29 October. The CME-driven shock arrived to the Earth at ∼2000 UT on 30 October and triggered sudden increase of the SYM-H index of ∼100 nT. The CME direct propagation velocity is estimated at about 1800 km/s.
 Fast CMEs propagating in the ambient solar wind (SW) generate strong foreword shocks, which enable effective acceleration of solar energetic particles (SEP) up to very high energies [Tylka, 2001]. Owing to continuous shock-driven acceleration, the SEP fluxes increase gradually and achieve maximum values at time of the shock arrival. This effect is clearly seen in Figure 1 at ∼0600 UT on 29 October and at ∼1000 UT on 30 October, when maximal fluxes are observed by GOES-10 for the SEP protons with energies up to >30 MeV. Note that the fluxes of high-energy protons observed at geosynchronous orbit are very close to that observed in the interplanetary medium [Cohen et al., 2001].
 Strong negative variations of the Dst and SYM-H indices at 0600–0900 UT and 1300–2400 UT on 29 October and at 1800–2300 UT 30 October indicate a very large magnitude and/or duration of the west-east component (Ey) of induced interplanetary electric field [Burton et al., 1975]. The Ey component is directly proportional to the SW speed and negative Bz component (in GSM) of the interplanetary magnetic field (IMF). Hence the solar wind speed and/or negative IMF Bz could be very large on 29–30 October 2003. High solar wind pressure and/or large negative IMF Bz causes such a strong magnetosphere compression that the magnetopause approaches and crosses geosynchronous orbit. Numerous geosynchronous magnetopause crossings (GMCs) are observed by GOES and LANL satellites during the considered interval [Dmitriev et al., 2005].
 The 29–31 October 2003 event was accompanied by extremely intensive fluxes of high-energy SEPs, very high solar wind pressure, i.e., very large solar wind velocity and/or density, and strong negative IMF Bz. Under such interplanetary conditions, space experimental data could be distorted due to following main reasons: saturation, energy range limitation, and SEP radiation effects. The saturation of space detectors is due to limitation of their dynamic range for measurements of the solar wind parameters. For example, the IMP-8 magnetometer range is limited by 35 nT [Luhmann et al., 1993], and thus the IMF with magnitude higher than 35 nT should be distorted in the IMP-8 measurements.
 The energy range limitation leads to missing a portion of data on solar wind plasma, which cannot be detected by instruments. The SOHO CELIAS/MTOF plasma monitor [Ipavich et al., 1998] and the Solar Wind Ion Analyzer in the Geotail CPI plasma experiment [Frank et al., 1994] detect ions from 0.3 to 6 keV/q and from 0.144 to 6.825 keV/q, respectively. This means that those instruments can only measure solar wind with speeds up to about 1000 km/s. In faster solar wind streams the SOHO CELIAS/MTOF and Geotail CPI underestimate the solar wind density because a significant portion of the plasma energy distribution occurs beyond the sensor range. The ACE SWEPAM detector is able to detect ions with much higher energy, up to 35.7 keV/q, in normal operation “track” mode [McComas et al., 1998] and thus it is able to measure very fast solar wind.
 Space data can be also distorted due to the radiation effects from very intensive fluxes of high-energy SEPs. On 28–31 October 2003 the SOHO coronograph was blinded by SEP and the ACE plasma instruments were overwhelmed by the blast and unable to provide reliable measurements [Lopez et al., 2004]. Namely, the high SEP background caused the ACE SWEPAM solar wind tracking algorithm to fail. The only valid solar wind measurements came from another operating “search” mode, which is run for approximately 1 min every 33 min. The search mode data are collected over a broad energy range from 250 eV/q to 17.9 keV/q with 10–12% energy resolution. Hence in the search mode the upper threshold of the measurable SW velocity for the ACE SWEPAM was about 1800 km/s.
 The SEP fluxes cause large backgrounds, degradation in sensors, and single event effects in microelectronics [Tylka et al., 1996]. Radiation effects in onboard equipment are caused mainly by nuclei having high enough energy for penetration through a satellite shielding. A typical satellite shielding of about 0.5 cm of aluminum is penetrated by protons with energy >30 MeV, which are able to produce a high level of ionization in electronic systems onboard a satellite. Hence enhancements of >30 MeV proton fluxes in the interplanetary medium might lead to increases of the space data distortions, such as data gaps and numerous errors.
 The saturation effect can be revealed easily as a cutoff in the data at some threshold. An estimation of the data distortion due to the energy range limitation or SEP radiation is more complicated and sometime unpredictable. In this sense a comparison of the SW data obtained from different satellites as well as an indirect estimation of the solar wind conditions would be helpful.
 In the present study we perform a comparative analysis of the SW conditions, which were observed by upstream monitors and estimated from the models. In section 2 we present experimental data on the solar wind plasma and IMF. To estimate the solar wind conditions, we use two independent data sets. The first one described in section 3 is based on time lags for propagation of solar wind structures from the ACE to Geotail satellites. The second data set is collected from GMCs observed by GOES and LANL satellites [Dmitriev et al., 2005]. The GMC data set is used for solar wind pressure estimations in section 4. In section 5 we demonstrate examples of the SEP radiation effects in the space plasma data. Section 6 is summary and discussion.
2. Solar Wind Data
 Solar wind conditions on 29–31 October 2003 were measured by the IMP 8, Geotail, SOHO, and ACE satellites. Wind satellite was located in the far tail of the magnetosphere or magnetosheath (GSM X ∼ −160 RE, Y ∼ −15 RE, and Z ∼ 4 RE) so we do not use its data. Solar wind plasma data from IMP 8 MIT experiment are retrieved from ftp resource belonging to MIT Space Plasma Physics Group (ftp://space.mit.edu/pub/plasma/imp/www/imp.html) with 1-min and 1-hour time resolutions. Geotail data on the solar wind plasma (CPI experiment) and IMF (MGF experiment) with ∼1-min time resolution are available on CDAWeb (http://cdaweb.gsfc.nasa.gov/). However, we use ∼1 min Geotail CPI data, in which the SEP background contamination is reduced (http://www-pi.physics.uiowa.edu/cpi-data/sw/). The IMF data from the ACE MAG experiment with 16-s resolution are presented on Web site http://www.srl.caltech.edu/ACE/. The ACE SWEPAM plasma data in the track mode are available on http://cdaweb.gsfc.nasa.gov/ and the search mode data are provided by R.Skoug (private communication, 2004). Solar wind plasma data from SOHO CELIAS/MTOF experiment with 1-hour time resolution are available on Web site http://umtof.umd.edu/pm/crn/.
 Measurements of the Geotail plasma wave instrument (PWI) [Matsumoto et al., 1994] are used to determine the electron concentration Ne in the solar wind. The electron concentration should be equal to proton concentration Np, if the contribution of α particles to the solar wind density is small. This is not true for some CMEs, which are characterized by large helium abundances, such that the Ne can be much (up to twice) larger than Np. On the other hand, the mass density (AMU per cm3) of the solar wind D is larger than the density estimated from the electron concentration because in helium and other heavier nuclei the number of nucleons is about two times larger than the proton number.
Figure 2 shows time profiles of the SW velocity and density measured by IMP 8, Geotail, ACE and SOHO on 29–31 October 2003. One can see a significant difference in the plasma data provided by the upstream monitors on 29–31 October. Note that at the considered time the ACE and SOHO satellites are located near point L1. The IMP 8 and Geotail satellites move very close to the Earth and they enter into the magnetosheath or magnetosphere during time intervals, respectively, from ∼1600 UT on 30 October to ∼1100 UT on 31 October and from ∼0400 UT on 30 October to ∼1100 UT on 31 October 2003. At that time the satellites detect relatively small plasma velocity and high (low) plasma density in the magnetosheath (magnetosphere). Hence the difference in the measurements can be only partially explained by difference in location of the satellites, especially IMP 8 and Geotail. Moreover, until 0600 UT on 29 October the upstream monitors provide similar SW plasma data. The best agreement one can find is for the SW velocity. However, the SW density detected by the ACE in the search mode is more than two times larger than detected by the SOHO, Geotail, and especially IMP 8.
 The most differences are revealed from 0600 UT on 29 October to ∼1300 UT on 31 October, when the interplanetary medium and magnetosphere are extremely disturbed. The Geotail CPI and SOHO detect very low density at 0600–2000 UT on 29 October. There are many gaps in the IMP 8 plasma measurements. The electron density restored from the Geotail PWI data is much higher (up to 2 orders of magnitude). The ACE data on the SW density are not available from 0600 UT on 29 October to 0400 UT on 30 October. The SW velocity measured by the ACE is extremely high at that time and reaches 2000 km/s or even more. From 2000 UT on 29 October to 0600 UT on 30 October all the upstream monitors observed the SW velocity of about 1000 km/s but the Geotail CPI and SOHO detected a smaller SW density than determined from the Geotail PWI experiment.
 Since ∼0600 UT on 30 October the ACE and SOHO satellites observe similar SW densities, excepting the time period from 1600 UT on 30 October to 1100 UT on 31 October, when ACE detects SW velocity enhancement up to 1700 km/s. At ∼0900–1600 UT on 30 October the IMP 8 and ACE satellites provide very similar plasma data. Relatively high densities and small velocities are observed by IMP 8 at ∼0300–0500 UT on 31 October in the magnetosheath. From 1100 UT on October 31 when IMP 8 and Geotail returned to the interplanetary medium, measurements of the SW velocity were very close for different upstream monitors. Note that at 1100–1400 UT on 31 October the Geotail PWI provides the electron density, which is practically equal to the proton density measured by the IMP 8 MIT and that density is several times larger than the proton density measured in the SOHO and ACE experiments. After 1400 UT the IMP 8 MIT, Geotail PWI, and ACE SWEPAM observe similar density profiles, while the SOHO CELIAS/MTOF and Geotail CPI experiments probably understate the solar wind plasma density.
Figures 3 and 4 present IMF components measured by ACE and Geotail, respectively, at 6000–7000 UT on 29 October and at 1530–1730 UT on 29 October. One can see a very good correspondence between the IMF data despite the large radial distance between the upstream monitors (>200 RE). The ACE time profile is delayed to achieve the best cross-correlation with the Geotail data. We have to emphasize that the high cross-correlation of the ACE and Geotail data on IMF exists for the entire interval on 29–30 October 2003. Hence the difference in the SW plasma measurements cannot be attributed to different solar wind structures observed by the satellites located in different places. There should be another reason for the inconsistencies in the SW plasma observations.
 If we suppose that ACE detects an actual solar wind velocity, then that fast solar wind is beyond the energy range of the IMP 8 MIT, Geotail CPI, and SOHO CELIAS/MTOF detectors, and thus they should underestimate the SW velocity and density. As a result, only the ACE SWEPAM in the search mode could provide reliable data on the solar wind plasma parameters and the Geotail PWI experiment allowed indirect estimation of the solar wind plasma density. Note that the ACE SWEPAM data have been corrected on the SEP background contamination. The correction algorithm might distort the actual data and, thus, the corrected data should be verified. The inconsistent measurements of the SW density at 0000–0600 UT on 29 October and at 1100–1400 UT on 31 October also should be verified.
3. Estimation of the Solar Wind Velocity
 To verify the experimental data of the solar wind plasma, we estimate the SW velocity from a time delay for direct solar wind propagation from the ACE to Geotail. At different moments we determine a time lag, which is required for the best cross-correlation between the ACE and Geotail data on IMF. In the estimation we also take into account tilted interplanetary fronts and the propagation of solar wind structures such as interplanetary shocks (IS), rotational discontinuities (RD), and Alfven waves (WS). The estimation results are presented in Figure 2 and Table 1. Here we assume radial solar wind propagation with velocity V = Vx and Vy = Vz = 0.
Table 1. Main Numbers for Estimation of the Solar Wind Velocity
 At 0611 UT on 29 October 2003 a strong interplanetary shock (IS) arrived at the Earth and initiated geomagnetic disturbances. Figure 3 shows time profiles of the IMF components measured by ACE and Geotail at 0600–0630 on UT 29 October 2003. At that time the magnetosphere was compressed by extremely strong solar wind pressure such that geosynchronous satellite LANL 1994-084 entered into the magnetosheath at ∼1600 LT [Dmitriev et al., 2005]. The low-latitude geomagnetic SYM-H index changes abruptly from −35 nT to −16 nT at 0611 UT and then gradually increases within 2 min up to −5 nT, also indicating a strong compression of the magnetosphere. The IS propagation time from ACE to Geotail is determined to be about dT = 12 ± 1 min, which testifies to very fast shock and the large speed of interplanetary transient generating the shock.
 The velocity observed by different satellites upstream of the shock varies about V ∼ 640 km/s. At 0451 UT on 29 October 2003 we identify a rotational discontinuity (RD), which propagates from the ACE to Geotail with the lag dT = 26 ± 1 min. Using the minimal variance method, we determine a wave normal for the RD as n = (0.81, −0.27, 0.52), i.e., its front is tilted at about 35°. The RD propagates in the solar wind along the normal n with Alfven speed Va. Hence the expression for the solar wind velocity vector V is the following:
where Δr is a vector of the Geotail and ACE relative locations Δr = rACE − rGeotail. To calculate the Alfven velocity, we estimate from the ACE magnetic measurements an average magnetic field for the RD of about Bk = 6.7 nT. Note that the IMF measurements on ACE and Geotail are very similar. However, a choice of the solar wind density is a problem. The best correspondence with the SW velocity of V ∼ 640 km/s measured by the upstream monitors can be achieved if we take the density of about D = 4 cm−3, which is mostly close to the electron density measured at 0517 UT by the Geotail PWI. Note that due to anisotropy of the SW thermal/magnetic pressure ratio β, Alfven speed for the RD could be smaller [Lin and Lee, 1994], and hence the estimated density also could be smaller down to 50%. Unfortunately, at the considered time, any information about the anisotropy is unavailable. Hence upstream of the shock the SW plasma density of about 2.6∼3.4 observed by SOHO and Geotail PWI, respectively, should be mostly reliable.
 The solar wind density and velocity downstream of the shock are ambiguous due to possible effect of energy range limitation for the SOHO and Geotail plasma detectors. We have estimated the upstream velocity V1 ∼ 640 km/s and density D1 ∼ 4 cm−3. From a magnetic coplanarity analysis of the ACE data the shock normal is determined as n = (0.68, −0.74, −0.03) and an angle between the shock normal and upstream magnetic field is found to be θBn = 23.5°. We consider a magnetic field jump on the shock to estimate the downstream density and velocity. From the ACE data the magnetic field upstream and downstream of the shock is obtained as B1 = 13.86 nT and B2 = 47.8 nT, respectively, and thus the magnetic field jump is equal to B2/B1 = 3.45. From this ratio one can estimate the IS Mach number of about Ma = 2.2∼2.5 for a quasi-parallel shock with θBn = 23.5°. The IS Mach number Ma = 2.3 and the upstream solar wind density D1 ∼ 4 cm−3 give the downstream solar wind density D2 ∼ 13 cm−3. Note that this density value is in very good agreement with the electron concentration of De ∼ 14 cm−3 obtained from the Geotail PWI experiment just behind the IS.
 The shock speed Vs is calculated from the time lag between the ACE and Geotail:
 Taking values of Δr, n and dT from Table 1 we find that Vs = 1030 km/s. In the shock frame the upstream solar wind velocity is equal to V*1 = Vs − (V*1n) = 615 km/s. In the shock frame the upstream (downstream) plasma velocity V*1 (V*2) and density D1 (D2) should satisfy the continuity condition:
from which we calculate the downstream velocity to be V*2 = 185 km/s. Taking into account the shock velocity and orientation, we find the SW velocity downstream the shock:
It is equal to V2 = 1260 km/s, which is in agreement with the ACE measurements. Hence downstream of the shock the ACE data of the SW velocity and Geotail PWI data on the solar wind density are reasonable.
 The solar wind velocity in the downstream region also can be estimated by using a tangential discontinuity (TD) at ∼0644 UT on 29 October 2003. A normal to the TD rotation plane is determined as n = (0.8, −0.6, 0.02). In case of the TD, frozen in the solar wind, the SW velocity vector V satisfies the following expression:
As a result we estimate the velocity of about V = 1600 ± 130 km/s, which is very close to the ACE measurements V = 1640 km/s in the downstream region at 0653 UT on 29 October 2003.
 Another example of the SW velocity estimation at 1530–1730 UT on 29 October 2003 by using structures TD and wave structure (WS) is presented in Figure 4. Taking data from Table 1 and equation (5) for the TD and equation (1) for the WS, one can calculate the SW velocity of about V = 1280 ± 100 km/s and V = 1120 ± 140 km/s, respectively. Note that the Alfven speed in equation (1) is calculated using the ACE data for the IMF and the Geotail PWI data for the electron density. An error of the velocity estimation is mostly due to the 1-min time resolution of the Geotail data, such that the lag can be estimated with no more than 1-min accuracy. The SW velocity measured by the ACE at corresponding time moments is V = 1370 km/s and V = 1354 km/s, respectively. One can see that the velocity estimated from the TD is in good agreement with the ACE data, while the WS gives slightly smaller SW velocity. Because the same IMF is detected firmly by both the ACE and Geotail satellites, the discrepancy in the velocity estimation might be due to an understated solar wind density, which is used for calculation of the Alfven speed of the wave structure. In order to adjust the velocity measured by the ACE, the SW density should be about D = 10 cm−3.
 Using the method described above, we estimate the solar wind velocity in the time period of 29–31 October 2003 (Figure 2). The tilted interplanetary fronts are taken into account during the time interval from 0600 to 1800 UT on 29 October 2003. From 0600 UT on 30 October to 1100 UT on 31 October the Geotail left the interplanetary medium. At that time we can estimate the time delay very roughly, using magnetic measurements on GOES-10 and GOES-12, when they enter into the magnetosheath. In general our estimation of the SW velocity is very close to the ACE observations, excepting a few cases. The largest discrepancies appear at 1600–1800 UT on 29 October, when the SW velocity is estimated from the wave structures and at 2100–2400 UT on 30 October, when the time lag is estimated occasionally and very roughly using GOES magnetic field data.
 As we mentioned above, the discrepancies might be caused by an incorrect determination of the Alfven speed for the WS that is due to the understated solar wind density. Adjusting of the estimated velocity to the ACE data requires the solar wind density of about D ∼ 9 cm−3 (instead 6 cm−3) at 1400–1423 UT, D = 10 cm−3 (instead 3.7 cm−3) at 1629–1653 UT, and D = 20 cm−3 (instead 4.6 cm−3) at 1740–1750 UT on 29 October. Note that the difference can not be explained by estimation error only. On the other hand, density-independent estimation of the SW velocity based on the TD at 1435 UT and 1552 UT on 29 October is consistent with the ACE data. Hence ACE should provide a reliable SW velocity. Only the time lag for solar wind propagation does not permit estimation of both the solar wind velocity and density. Another independent way for determination of the solar wind plasma characteristics is required.
4. Estimation of the Solar Wind Pressure
 An estimation of the solar wind pressure in this section is based on application of magnetopause models for a data set of geosynchronous magnetopause crossings (GMCs) observed by GOES-10, GOES-12, LANL satellites 1990-095, 1991-080, 1994-084, and LANL-97A on 29–31 October 2003 [Dmitriev et al., 2005]. We use a magnetopause model by Kuznetsov and Suvorova  (hereafter KS98 model), a modified Chao's model [Chao et al., 2002; Yang et al., 2003] (hereafter Ch03 model), and a model by Shue et al.  (hereafter Sh98 model). The models permit calculation of the SW pressure required for a GMC at given direction, which is characterized by GSM longitude and latitude, and for given IMF Bz in GSM. We use Geotail data on IMF Bz during time period from 0600 UT to 2400 UT on 29 October and ACE data in 30 and 31 October.
Figure 5 shows the SW dynamic pressure Pd estimated from the magnetopause models. The lower level of the pressure (black circles) is obtained for each 5-min as a maximum of the solar wind pressures calculated by a model for one or several geosynchronous satellites located in the magnetosheath. In other words, the lower level corresponds to the pressures, which are required for geosynchronous satellites situated mostly far from noon to be located in the magnetosheath. The upper level of the Pd (gray circles) is obtained for each 5-min as a minimum of model predictions of magnetopause crossings for geosynchronous satellites, which are located inside the magnetosphere. The upper level corresponds to the pressures, which could produce magnetopause crossings for geosynchronous satellites located in the magnetosphere mostly close to noon. We choose the 5-min duration because of difference in time resolution and in moments of measurements for different satellites. The upper and lower dynamic pressures constrain a “corridor” of possible values of the SW dynamic pressure, which is required for a model explanation of the observed GMCs.
 Apparently, the upper pressure level (satellites are still in the magnetosphere) should be higher than or equal to the lower level (satellites are already in the magnetosheath). In Figure 5 one can find several cases of violation of that rule. Some of those cases are associated with such fast and large variations of the IMF Bz that within a 5-min interval the maximum pressure required for the magnetosheath interval appears larger than the minimum pressure required for the magnetosphere interval. However, those variations take place only in a few cases, at ∼0600–0900 UT on 29 October, ∼1600–1800 UT on 30 October, and at 0000–0200 on 31 October.
 Another reason of the violation is a prominent dawn-dusk asymmetry of the magnetopause, which is appearing as a duskward skewing of the dayside magnetopause during the main phase of geomagnetic storms [Dmitriev et al., 2004, 2005]. For the asymmetrical magnetopause, the symmetrical models Sh98 and Ch02 overestimate the SW pressure required for magnetopause crossings in the prenoon sector and underestimate it in the postnoon sector. The KS98 model predicts the skewing as a function of the IMF Bz. However, the relationship between the skewing and Bz is ambiguous [Dmitriev et al., 2005]. As a result, the KS98 model can underestimate the SW pressure required for magnetopause crossings in the prenoon sector and vice versa. The overestimation/underestimation of the pressure leads to contradictory predictions, when the lower level is higher than the upper level of the pressure.
 Note that the Sh98 and Ch03 models take into account a helium contribution into the dynamic pressure that is usually accepted to be equal to its average value of 4%. For most GMCs on 29–31 October the He contribution is unknown; hence we divide the pressure in the Sh98 and Ch03 models by 1.16 in order to obtain a “pure” dynamic pressure of the solar wind protons. The shifting parameter dY in the KS98 model is accepted to be dY = 0.3 REthat permits minimization of contradictory predictions (the upper pressure is smaller than the lower one), which are caused by the model overestimation of the magnetopause dawn-dusk asymmetry [Dmitriev et al., 2005].
 For comparison in Figure 5 we show a combined solar wind pressure Pd (black solid line), which is calculated using the SW velocity from the ACE SWEPAM data and electron density from the Geotail PWI data. During the time interval from 0600 UT on 30 October to 1000 UT on 31 October, when the Geotail is located in the magnetosheath, the solar wind dynamic pressure is calculated using the ACE SWEPAM data (gray solid lines). The helium contribution is available only on 31 October and it is taken into account. Estimation of the SW pressure is performed also using the time lag for propagation of the IS, RD, and WS structures (crosses). For this purpose we use the SW velocity provided by the ACE and densities, which were estimated in the previous section to adjust the ACE SWEPAM velocity profile.
 In case of perfect agreement, the calculated SW dynamic pressure should lie just inside a corridor restricted by the modeled lower and upper pressures. We note a good agreement between the Pd estimation based on the time lag and predictions of the magnetopause models, especially for the KS98 model. The solar wind dynamic pressure obtained from the ACE SWEPAM and Geotail CPI experiments is mainly smaller than the predictions of the Sh98 and Ch02 models. The KS98 model demonstrates a good correspondence to the experimental data more often.
 To simplify a comparative analysis, we combine the model predictions and calculate a joint corridor for the SW pressures required for GMCs. In the joint corridor the lower and upper pressures are calculated for each 5-min interval from the KS98, Sh98, and Ch03 models. Examples of the calculation are presented in Table 2. For a satellite located in the magnetosheath (magnetosphere) we take the minimum (maximum) of the pressures predicted by the different models. Then, among the satellites located in the magnetosheath (magnetosphere), we choose maximum (minimum) of the taken minimal (maximal) pressures. Hence the lower and upper boundaries of the joint corridor can be considered as the absolute minimum and maximum dynamic pressure, which can be predicted from the magnetopause models. From the pressure corridor we estimate a corridor for the solar wind density, using SW velocity provided by the ACE SWEPAM. By this way, we obtain an estimation of the SW proton density because we have already eliminated the helium contribution from the model predictions.
Table 2. Solar Wind Dynamic Pressure Measured and Calculated From the MP Models
 The estimated and measured SW density and pressure are presented in Figure 6. One can see that the indirect estimations are more or less close to the observations at ∼0600–1600 UT and at 1800–2400 UT on 29 October, at 0000–0400 UT and at 1900–2400 UT on 30 October, and at 1100–1300 UT on 31 October. In the latter case a very good agreement is revealed only with the Geotail PWI data. Note that at 0500–0700 UT on 31 October the Geotail, located mainly in the magnetosheath, observed multiple bow shock crossings (see Figure 2), that also indicates a strong compression of the magnetosphere. On 29–30 October the ACE SWEPAM operated in the search mode and provided the plasma data with ∼33 min time step. Owing to this the ACE SWEPAM could miss <30 min enhancements of the SW density at 1900–2400 UT on 30 October.
 Several cases of comparison between the SW pressure estimations and the ACE and/or Geotail observations are presented in Table 2. The pressure PdPWI is calculated from the electron concentration provided by the Geotail PWI and from solar wind velocity VACE measured by the ACE SWEPAM. The helium contribution, when available, is taken into account. One can see that in most cases the estimations and PdPWI are in very good agreement. On the other hand, the pressure PdACE provided by the ACE SWEPAM is much lower than the joint corridor for the SW pressures required for GMCs. Moreover, at 1135 UT on 31 October the PdACE is about twice less than the PdPWI, which is well supported by the IMP 8 measurements (see Figure 2).
 An underestimation of the density and pressure by the Geotail PWI can be found at 1600–1800 UT on 29 October. It might be due to low-quality data, which contain many gaps at that time. The ACE SWEPAM systematically underestimates the density and pressure at 1700–1800 UT on 30 October and at 0000–1300 UT on 31 October. Note that at 0400–1000 UT on 31 October the magnetopause response on the solar wind conditions is ambiguous because of Pc-5 global mode magnetospheric pulsation [Dmitriev et al., 2005]. The indicated above intervals are accompanied mainly with positive IMF Bz. Hence very high solar wind pressure of Pd > 20 nPa is required for the observed GMCs [Suvorova et al., 2005]. As one can see in Figure 6, large enhancements of the Pd should take place at ∼1100–1800 UT on 29 October, at ∼1700 UT on 30 October, and at ∼0000–1300 UT on 31 October, when the SYM-H index has large positive variations, which testifies to strong compression of the magnetosphere.
 We find that the model estimation of the solar wind pressure and density is in reasonable agreement with the ACE SWEPAM and especially with Geotail PWI measurements, excepting three intervals at 1600–1800 UT on 29 October, 1700–1800 UT on 30 October, and 0000–0400 UT on 31 October. In that time the differences are significant such that the estimated magnitudes of the solar wind pressure and density are much larger (up to 10 times) than observed ones. Note that the case at 1600–1800 UT on 29 October is supported by two independent estimation methods. The discrepancies on 30 and 31 October are revealed only with predictions of the magnetopause models and thus they require additional verification.
5. SEP Radiation Effect
 As we mentioned in section 1, the 29–31 October 2003 event is accompanied by very large fluxes of high-energy SEP. The SEP-associated contamination is easily revealed in the Geotail Solar Wind Ion Survey spectrogram on 28–30 October 2003 (Figure 7). During the time intervals from ∼1200 UT on 28 October to ∼1200 UT on 29 October and from ∼2200 UT on 29 October to ∼1800 UT on 30 October, one can see a very intensive broadband energy-independent background, which comes from interaction of the SEP with the sensor. This so-called “white noise” contributes a lot to the integral of the plasma density. A restoration of the actual solar wind plasma data requires the development of special methods, which permit accurate subtraction of the radiation background. Hence the reliability of the solar wind plasma data during strong SEP events depends directly on the accuracy of the method used for elimination of the SEP contamination.
 To investigate the reliability of methods used for restoration of the space plasma data, we compare time profiles of the SW velocity and density measured by different satellites during two case events, which are accompanied with very intensive SEP fluxes. An event on 8–10 November 2000 is presented in Figure 8. The event can be divided on two intervals. During the first one on 9 November the solar wind plasma data are distorted dramatically by a very intensive SEP flux such that the Geotail, Wind, SOHO, and ACE plasma measurements are much different. The ACE SWEPAM turns into the search mode. There is a long gap in the Wind data on the SW velocity and Geotail data on the solar wind density. It is rather difficult to estimate the data validity at that time. Note only that Wind is located very far on flank (X ∼ 80 RE, YZ ∼ 80 RE).
 Since 1800 UT the SEP flux decreases down to <103 (cm2 s sr)−1 and IMP-8, Wind, SOHO, and ACE data are matched very well. On 10 November the satellites IMP-8, Wind, and SOHO demonstrate excellent agreement in measurements of the SW density, while the densities detected by Geotail and ACE are smaller. The most prominent difference can be found at 0600–0900 UT on 10 November, when the solar wind velocity increases up to 900 km/s and the >30 MeV SEP flux is about 100 (cm2 s sr)−1. Multiple GMCs observed by LANL satellites at that time requires the SW pressure of about 20 nPa or even more because the GMCs are observed at ∼1400 LT under positive IMF Bz (not shown). However, ACE provides smaller SW pressure and density. Hence at 0600–0900 UT on 10 November 2000 we find two independent sources of evidence that the SW density was underestimated in the ACE SWEPAM experiment.
Figure 9 presents SW plasma measurements from IMP-8, Geotail, Wind, SOHO, and ACE during the Bastille event on 14–16 July 2000. Before the SEP enhancement at ∼1200 UT on 14 July all the satellites demonstrate very good agreement in measurements of the plasma parameters. Note that Geotail is located in the outer magnetosphere and magnetosheath until ∼1600 UT on 15 July. During the SEP enhancement Wind data of the SW velocity are not available and ACE SWEPAM turns into the search mode. Note a perfect agreement in the SW velocity measured by IMP-8, Geotail, and ACE. At 1200–2400 UT on 15 July, when the solar wind velocity exceeds 900 km/s and >30 MeV SEP flux is larger than 50 (cm2 s sr)−1, the SOHO and ACE SWEPAM provide the SW density, which is much smaller than the density measured by IMP-8, Geotail, and Wind. At the same time, GOES and LANL geosynchronous satellites observe long-lasting magnetosheath intervals in a wide range of local times. The magnetopause models predict the SW pressure of >20 nPa that is in agreement with the IMP-8, Wind, and Geotail observations. Since ∼0200 UT on 16 July, when the SEP flux is less than 50 (cm2 s sr)−1 and SW speed is lower than 1000 km/s, all the satellites (IMP-8, Geotail, Wind, SOHO, and ACE) detect practically the same solar wind density. As one can see, the Bastille event scenario is very similar to the 29–31 October 2003 event.
 We find that the SW density and pressure provided by the ACE SWEPAM can be understated during disturbed intervals, when the solar wind speed is higher than 900 km/s and integral fluxes of the SEP with energy >30 MeV are more intensive than 50 (cm2 s sr)−1. Moreover, larger SEP fluxes and faster solar wind cause larger distortions, right up to the data losses. Note that consideration of only the integral flux of SEP with energy >30 MeV is rough simplification. For higher accuracy one should regard a spectrum of high-energy particles in the interplanetary medium.
6. Discussion and Summary
 Comparison of the solar wind data provided by upstream monitors SOHO, ACE, and Geotail on 29–31 October 2003 give ambiguous results. On the one hand, we find a high correlation for the IMF measured by ACE at the L1 point and by Geotail in close vicinity of the Earth that is in good agreement with previous studies of the IMF correlations [e.g., Richardson and Paularena, 2001]. The high correlation proves reliability of the IMF data provided by both the ACE and Geotail satellites in the present case. On the other hand, data on the SW velocity and density provided by different satellites are much different. Our estimation, based on the time lag for propagation of the SW structures from the ACE to Geotail, shows that the most reliable velocity is provided by the ACE SWEPAM experiment. However, in some cases, the estimated velocity is much smaller than the ACE measurements. Adjusting the estimated velocity to the ACE data requires SW density values larger than measured by the upstream monitors. This disagreement might be attributed to the density underestimation in the space plasma experiments at 1600–1800 UT on 29 October.
 An application of the KS98, Sh98, and Ch03 magnetopause models for the geosynchronous magnetopause crossings observed by the GOES and LANL satellites on 29–31 October enables us to estimate the solar wind dynamic pressure. The SW pressure corridor corresponds well to the Geotail PWI and ACE SWEPAM measurements, excepting three intervals at 1600–1800 UT on 29 October, 1700–1800 UT on 30 October, and 0000–0400 UT on 31 October.
 The large difference between the model estimations and experimental data could indicate the inability of the models to predict the strongly disturbed magnetopause. However, previous studies [Shue et al., 2000, 2001; Yang et al., 2002] convincingly show that the magnetopause models have a high capability for the GMC prediction. The KS98 and Ch03 models were developed using data sets containing several GMCs. Observations of the GMCs on 31 March 2001 [Ober et al., 2002] show that the magnetopause models successfully predict the dayside magnetopause for extremely compressed magnetosphere when a portion of geosynchronous orbit occurs in the interplanetary medium. Note that on 31 March 2001 the intensity of >30 MeV protons was less than 1 (cm2 s sr)−1 and therefore the SW data could not be distorted by the SEP.
 On the other hand, a comprehensive analysis of the solar wind conditions for hundreds of magnetosheath intervals [Suvorova et al., 2005] shows that the models overestimate the SW pressure required for GMCs, especially for northward IMF. Hence the relatively large SW dynamic pressure predicted by the models can be due to the model shortcoming. However, the good correspondence of the model predictions with estimations based on the time lag for the SW propagation at 0600–1800 UT on 29 October testifies to the reliability of the model prediction even for northward IMF. Moreover, the solar wind density estimated from the models is in very good agreement with the electron density provided by the Geotail PWI experiment. Note that here we suppose that the ACE SWEPAM provides reliable solar wind velocity.
 It is important to emphasize that the estimation methods of the magnetopause models and the time lag for the SW propagation are opposite in a sense of adjusting the solar wind density. The method of magnetopause models allows estimation of the solar wind dynamic pressure required for GMC. For given pressure, a smaller density should be estimated for higher velocity. The method of time lag is based on equation (1), which contains a sum of the Alfven speed and the parallel component of the solar wind velocity. In that expression, for higher velocity the adjusted density should also be higher to reduce the Alfven speed. In other words the solar wind density and velocity estimated using the two methods are self-consistent. Hence we can suggest that the solar wind velocity provided by the ACE SWEPAM should be reliable. The inconsistencies in the SW velocities and pressures are due to the understated solar wind density.
 In section 5 we demonstrate two case events, in which the SW density provided by the ACE SWEPAM is understated under strongly disturbed conditions. Those conditions are characterized by very fast solar wind (>900 km/s) and intensive fluxes (>50 (cm2 s sr)−1) of the SEP with energy >30 MeV. The same conditions are observed on 29–31 October 2003. As one can see in Figure6 at 1700–1800 UT on 30 October and 0000–0400 UT on 31 October, the solar wind speed is much higher than 900 km/s and the SEP flux decreases from ∼50 to 10 (cm2 s sr)−1. Under such conditions the ACE SWEPAM can provide significantly understated solar wind density, much smaller than that required for the GMC prediction. Hence a method of the plasma density restoration from the ACE SWEPAM data is unreliable for very disturbed heliospheric conditions such as on 29–31 October.
 Accounting for the helium contribution reduces the inconsistencies only partially. Note that since ∼0100 UT on 31 October the α/p ratio used for the solar wind pressure calculation is based on the ACE SWEPAM data. However, at 0000–0400 UT on 31 October the solar wind pressure calculated with the helium contribution is still much lower (few times) than the model estimation. The large helium abundant inside the CME in the second half of 29 October could partially explain disagreements between the estimates and Geotail PWI data on electron density, especially at 1600–1800 UT. Besides the Geotail data have many gaps at that time, thus large enhancements of the density could be missed. At 1700–1800 UT on 30 October the model estimation is about 10 times larger than the pressure provided by ACE SWEPAM. Such a big difference can not be explained by the large He abundance. Otherwise the helium contribution would be impossibly large (hundreds of percents).
 Finally, we can interpret the solar wind data provided by different upstream monitors from 0600 UT on 29 October to 1400 UT on 31 October 2003. The ACE and Geotail provide reliable data on the IMF. Reliable data on very high solar wind speed is restored from the ACE SWEPAM measurements. Owing to the very high solar wind speed the IMP 8 MIT, Geotail CPI, and SOHO CELIAS/MTOF detectors underestimate the solar wind density and velocity. Until 0600 UT on 30 October the solar wind density can be obtained from the Geotail PWI data on the electron concentration. Note that at 1600–1800 UT on 29 October the Geotail PWI data quality is low such that the solar wind density might be understated. Since 0600 UT on 30 October the ACE SWEPAM provides data on the solar wind density, which is reliable only partially. From ∼0800 UT to ∼1600 UT on 30 October, when the solar wind velocity was about 1000 km/s, both the IMP 8 MIT and ACE SWEPAM observed very similar and thus reliable solar wind velocity and density. A significant underestimation of the density is revealed at 1700–1800 UT on 30 October and 0000–0400 UT and 1100–1400 UT on 31 October. Actual solar wind density should be at least 2 to 3 times larger than restored from the ACE SWEPAM data. Namely, the significant enhancement of the solar wind density up to ∼50 cm−3 is correctly observed at 1100–1400 UT on 31 October only by the IMP 8 MIT and Geotail PWI experiments.
 The authors thank NASA/GSFC ISTP for providing data from the Geotail, Wind, ACE, GOES, and LANL satellites. We thank T. Nagai from Tokyo Institute of Technology Earth and Planetary Sciences for providing the Geotail magnetic data. We also thank collaborative efforts of GSFC, UNH, MIT in providing the Wind plasma data, C. W. Smith from the University of New Hampshire for providing the ACE magnetic data, MIT Space Plasma Physics Group for providing IMP 8 plasma data, and Kyoto World Data Center for Geomagnetism for providing the Dst and ASY/SYM indices. Authors are very grateful to R. Skoug from Los Alamos National Laboratory for providing the ACE SWEPAM data and invaluable discussion of the plasma measurement method. This work was supported by grants NSC92-2811-M-008-021 and NSC-92-2111-M-008-003.
 Arthur Richmond thanks Ramon Lopez and another reviewer for their assistance in evaluating this paper.