Human space exploration is expected to enter its next phase in the coming decades as the United States prepares to return to the Moon or perhaps venture even further with a crewed mission to a near-Earth asteroid. Both mission classes are viewed by NASA as precursors of eventual crewed missions to Mars. In anticipation of extensive robotic and human presence in the space environment beyond the protection of the Earth's magnetosphere, it is important to better quantify and bound effects of earthward directed solar storms not just on the human body but also on engineering signals. In this paper, we study the effects of solar storms on S band (∼2.3 GHz) radio links in the near-Earth environment, primarily for application to navigation. In particular, we are concerned with induced long-period signatures on Doppler tracking data that could be confused with the Earth's gravity signature, resulting in perturbed trajectory solutions of returning spacecraft during Earth entry targeting. We have quantified “worst-case” levels of such induced signatures on S band signal phase using model predictions based on measured in situ charged particle content from satellites and have compared these results with signatures seen in actual tracking data during periods of interplanetary coronal mass ejections (ICME) and related geomagnetic storms. We show that induced Doppler can mask Earth gravity field effects in navigation trajectory solutions at S band, a commonly used frequency for near-Earth communications and navigation. Finally, we suggest a few ways that such effects can be identified, alleviated or eliminated in near real-time.
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.
 NASA is developing plans for human crewed missions that will return to the Moon, venture beyond to a near-Earth asteroid, and eventually land on Mars. In anticipation of future robotic and human presence in the near-Earth environment, and beyond, there is a need to quantify and bound the effects of earthward directed solar storms on the human body as well as on signals carrying navigation and telemetry information. Such information will be a safety-critical factor in successfully returning crew to Earth, especially along the high-energy return trajectories associated with asteroid or Mars missions. Here we define the solar storm as solar activity phenomena in the near-Earth environment and the Earth's ionosphere caused by flares and related Interplanetary Coronal Mass Ejections (ICMEs). Assessing solar storm effects on the human body have become one focus of such preparation activities given that the resulting prolonged radiation effects are a concern. As a result, solar storm studies (ICME and flare effects specifically) are being conducted to allow predictive warnings such that astronauts, who may be walking around on the surface of the Moon, can return to the protective shielding of their vehicles or bases during such events [Schwadron et al., 2007, 2010; Cucinotta et al., 2010].
 The availability of improved predictions is also important for minimizing impact on deep space communications [Woo, 2006]. The focus of this paper is to assess the effects of earthward directed solar storms, being manifest as ICMEs, on S band radio links in the near-Earth environment for application to navigation and telemetry, using model estimations and available data analysis. In particular, we are concerned with effects on the received Doppler, especially those with long-period signatures that could, to some extent, be confused with the Earth's gravity signature by navigation filters and result in perturbed Earth-return atmosphere entry targeting, with possibly serious consequences.
 Coronal mass ejections (CMEs) are massive bursts of material consisting of charged particles that are expelled outward from the Sun [Klein and Burlaga, 1982; Tsurutani and Gonzalez, 1997]. These events tend to be associated with large solar flares. A CME consists of a fast moving shock, a sheath (which is formed from shocked heated slow solar wind), and a magnetic cloud (see for instance a review by Tsurutani and Gonzalez [1997, and references therein]). In complex cases of multiple flaring, multiple structures will be formed. High density plasma associated with CMEs is found in sheaths. Magnetic clouds within CMEs are characterized by low steady density and large magnetic fields. If a southward magnetic field component is present in a magnetic cloud it interacts with the Earth's magnetosphere through magnetic reconnection and causes magnetic and ionospheric storms. The intensity of the geomagnetic disturbances is determined by the duration and magnitude of southward turning of magnetic field in the interplanetary structure [Tsurutani et al., 1999]. If both sheath and magnetic cloud have southward magnetic fields, a double magnetic storm occurs. A northward-oriented magnetic field in the structure will not cause a geomagnetic storm, but a shock could produce some geomagnetic activity.
 CME speeds can reach up to 2000 km/sec [Cliver et al., 1990]. When earthward directed, ICMEs reach the Earth's orbit in less than a day to several days later. Large ICME-type magnetic storms are typical during solar maximum years and can cause satellite communication disruptions, power grid outages, and other phenomena [National Research Council, 2008; Feynman and Gabriel, 2000].
 There are two interplanetary phenomena which can cause sudden and significant changes in ionospheric density during a solar storm. Extreme solar flares are known to result in ionospheric disturbances due to increased EUV radiation fluxes [see Prölss, 2004, and references therein]. Very strong flares occurred on 4 November 2001, 28 October 2003, and 29 October 2003 and have been compared and contrasted to the 14 July 2000 Bastille Day event [Tsurutani et al., 2005, 2009]. The peak TEC enhancement on 28 October 2003 was about 25 TECU (1 TECU = 1016 particles/m2) [Tsurutani et al., 2005] or about 30% above the background level, and lasted ∼3 h, significantly exceeding the duration of the solar flare. Enhancements in TEC observed during the 4 November 2003 event were of order ∼7 TECU (with ∼102 TECU background). On 29 October 2003, the enhancement was ∼5 TECU relative to a ∼90 TECU background) and for Bastille Day, the enhancement was about ∼5 TECU (with background of about ∼69 TECU) [Tsurutani et al., 2005]. The 28 October 2003 event was the most intense of the four when measured in the EUV [Tsurutani et al., 2005]. In this paper we will discuss effects in Doppler residuals which correspond to relative column densities larger than 20 TECU for time intervals longer than three hours. Thus, we will focus on non-flare causes of such effects.
 The arrival of the ICME itself (if it is geomagnetically active) occurs about a day or so after a strong flare which provides another source of ionospheric density enhancements.
 There are two contributions to the disturbed Doppler on the signal link due to an incoming solar storm to be discussed and compared in this paper: the effect of the ICME structure itself on the signal link and the secondary effect of a disturbed ionosphere. In addition to concerns with navigation solutions, storm associated perturbations of the signal phase may affect radar and other systems, and there may also be associated signal amplitude perturbations including signal loss [Feltens et al., 2004a, 2004b]. The latter could affect S band links in a variety of ways.
 Beyond the ionosphere, within the near-Earth environment, there are several spacecraft that possess instrumentation to measure in situ quantities such as charged-particle number density. Such measurements, acquired during earthward directed ICME events, can be used to quantify and bound contributions to signal phase (Doppler). Such spacecraft include those located upstream of the solar wind at L1. Dual-frequency GPS data can be used to quantify the secondary ionospheric component.
 In this paper, we calculate the Doppler shift from both the ICME components and the ionospheric components and compare these to S band Doppler tracking data from the Genesis spacecraft, which were routinely acquired during its mission, and included periods coinciding with Earth arrival of known ICME events. The levels of induced signatures seen in the Genesis Doppler data were compared with model predictions to better understand their effects on navigation solutions, and to characterize the corresponding velocity errors, and the time scales of these variations.
 We will provide theoretical background for associating changes in relative electron number density along the signal path with signal phase fluctuations (section 2), quantify effects on S band signal phase with electron content both over the signal path and due to the disturbed ionosphere (section 3), present the data selection criteria and the observations studied in this paper (section 4), discuss analysis of solar storm induced signatures found in selected S band tracking data during arrivals of the known ICMEs (section 5), and finally, propose mitigation strategies on how such effects can be alleviated or eliminated in near-Earth spacecraft trajectory solutions (section 6).
2. Theoretical Background
 As already discussed, there are two principal contributions of degradation to spacecraft signal reception due to ICME events. These are (1) the increased interplanetary electron density due to an ICME passage, and (2) the increased ionospheric density in the course of a solar storm. Both contributions impact signal phase to varying degrees. The first contribution is attributed to charged-particle number density (Ne) variations along the signal link between transmitter and receiver (excluding ionosphere) as the solar storm transits through the Earth-Moon environment. The second contribution is due to variation in total electron content (TEC) in the ionospheric layer that is transited by the signal during a solar storm. It is well known that geoeffective ICMEs produce convective electric fields that can partially penetrate into the ionosphere causing ionospheric uplift and increase in daytime TEC [Kelley et al., 2003; Tsurutani et al., 2004, 2008]. This effect depends on the strength of the ionospheric storm, geographical location of a receiver and local time of the event, as well as the look (or elevation) angle through disturbed regions of the ionosphere. Such effects may be more significant for a station located under the Equatorial Ionospheric Anomaly (EIA) or near a pole, while they may be at a much lower level for a signal path through the midlatitude ionosphere. However, an incoming solar super-storm can lift the dayside EIA to higher latitudes leading to enhanced density at 30 to 35 deg latitudes, thus affecting midlatitude TEC. Whereas on the night side, the storm will lower the ionospheric density thus reducing ionospheric TEC [Tsurutani et al., 2004, 2008; Mannucci et al., 2005; Verkhoglyadova et al., 2008]. Below we will analyze contributions to Doppler fluctuations in a signal link during a solar storm from ionospheric and interplanetary components separately.
2.1. Ionospheric Impact
 For the case of an ionospheric layer, we can model the signal phase fluctuations in terms of a total mean square phase fluctuation, 2 (in radians) which is in turn related to other quantities of the form [Flock, 1987]
where re is the classical electron radius (re = μ0q2/(4 π me) = 2.82 × 10−15 m), D is the Fresnel zone size of the irregularities (m), λ is the signal wavelength (m), L is the thickness of the ionosphere or plasma irregularity layer, is the local elevation angle, and is the electron number density fluctuation squared (m−6) that characterize the irregularities in the vertical direction. In addition, μ0 is the permeability of free space (4π × 10−7 H/m), q is electron charge (1.6 × 10−19 Coulombs) and me is electron mass (9.1 × 10−31 kg). The expression (1) is applicable to the ionosphere as it has terms describing thickness and angle for a radio link through the near-Earth layer.
 We are only interested in large-scale density fluctuations that induce long period trends into the data since they affect the trajectory solutions. Rapidly fluctuating short-term effects tend to be zero mean and reach higher peak frequencies. Consequently, they will not induce biases into the data.
 A simplified form of (1) at the exit plane of the signal path in units of cycles is given by
where δNe is the electron number density fluctuation (m−3) integrated through the irregularity layer of thickness L, f is the link frequency in Hz and c = 3 × 108 m/s [Rino, 1980]. In general, the fluctuation in phase can be related to δTEC (m−2), the fluctuation in TEC resulting from the variation of the integration of the electron number density Ne(l) along a signal path of length L [Flock, 1987]. Thus we now define δNe and δTEC along line of sight and thus elevation angle dependence (1/sinθ) is folded in.
 During transit through the ionosphere, radio signals will be scattered. Weak scattering is applicable at S band for the case of the ionosphere. For characterizing phase effects on signals (equation (1)), we assume Fresnel scale sizes of D = 800 km, and a thickness of L = 200 km for the ionospheric shell where the vertical height to the center of the ionosphere is h = 300 km [Flock, 1987]. During periods of high solar activity, h can reach higher values up to and perhaps above ∼450 km as used in multishell models such as those described by Hernández-Pajares et al. .
 The column density TEC in units of number/m2 is defined as follows:
 If we consider the worse case fluctuation in electron number density as the value of the peak electron number density over the entire link distance, then the fluctuation in phase can be bounded as follows:
 Here, we have a modified version of (2) using (3) over a link distance of L such that
 The Doppler fluctuation (in Hz) can be related to the phase fluctuation (in cycles) given in (2) or (4) and the count time T as
 The resulting velocity fluctuation (assuming two-way Doppler) is
2.2. Interplanetary Component
 As ICMEs expand outward from the Sun, the electron densities generally tend to decrease and the plasma piles up in a sheath ahead of it. Some plasma will be lost during the progress of a ICME either over the top, along its sides and by radial expansion [Tsurutani et al., 1988]. The densities of ICMEs at the Earth orbital distance of 1 AU can be measured by a number of satellites at the “gravitationally balanced” L1 libration point and elsewhere in the near-Earth vicinity. The electron density signature over the link inferred from such measurements can be used in (3) to obtain estimates of TEC. Alternatively peak values of electron number density measured by these satellites during a solar storm event can be used directly in (5) to achieve an upper bound for the interplanetary component of δTECmax. This value can then be input to (4) and thus (6–7) to quantify an upper bound for the interplanetary component of the Doppler frequency or velocity.
2.3. Estimation of TEC Changes from Tracking Phase Measurements
 We can estimate the relative TEC signature from the observed phase measurements by making use of the following formulation:
where i = 1 to n is the data point index of the Doppler time series, TEC(t0) is defined by anchoring the first phase point of the time series to the TEC of the quiet background prior to the advent of the ICME (since Doppler is only sensitive to changes in TEC). If the ICME is already in progress, the anchor point must be a point in the time series known to be quiet. The quiet background level can be extracted from an electron number density model or satellite measurements made at ∼1 AU. If the spacecraft has a particle counting instrument, then the time series can be compared with the radio sounding time series.
 For ionospheric disturbances, measurements of δTEC and TEC are usually available from global ionospheric network measurements. TECs from multiple satellite-to-ground links can be mapped to the signal path of interest through the ionosphere which in turn can be input into the relevant model to assess the ionospheric disturbance contribution for a particular site [Mannucci et al., 1998].
 For the region lying beyond the ionosphere, measurements of electron number density are available from several spacecraft which can then be converted to TEC along a signal path under certain model assumptions (see below). The case where the signal path distance L traverses half of the Earth-Moon distance of 406,000 km (L = 203000 km) is a crucial point for a mission scenario involving humans where the acquired phase data would be used in a trajectory solution in preparation of Earth re-entry targeting.
3. Expected TEC Levels and Effects on Signal Phase Due to Solar Storm Events
3.1. ICME Effects in Near-Earth Environment Excluding the Ionosphere
 Typical electron number densities due to ICMEs at a 1 AU Earth-Sun distance reach values of 10–100 cm−3 [Fry et al., 2004; Mulligan et al., 1999]. These levels can be contrasted with typical background levels during non-storm conditions of ∼6 cm−3 at 1 AU. A comparison of electron number density measurements between the NEAR and WIND spacecraft was performed by Mulligan et al. . Measured pre-shock levels ranged from 8 to 12 cm−3 and the post-shock levels over several events ranged from 29 to 79 cm−3. WIND/Waves thermal noise receiver data exhibited electron number density values of up to ∼40 cm−3 between 1994 and 2003 [Issautier et al., 2005]. For the purpose of our analysis, we adopt a level of 100 cm−3 as a representative upper bound for the fluctuation in electron number density in an ICME sheath during extreme events (δne,max ∼ ne,max), and 10 cm−3 for a nominal density in the ICME and interplanetary environment.
 For a peak electron number density fluctuation of 100 cm−3 over a distance of halfway to the Moon, the corresponding worst-case fluctuation in Doppler is about 0.06 Hz over a 60 s count time (using section 2 formulation). The corresponding velocity variation is 8.7 mm/sec (using equation (7)). This time-rate-of-change is comparable to an acceleration or force due to Earth gravitational field variations and can thereby “fool” a navigation filter. The time-scale for such fluctuations will of course depend upon the dynamics and density structure of the particular ICME as it passes through the signal link. For nominal events, the effects are about ten times lower. In practice, a sustained level of 100 cm−3 over the entire link distance is unlikely and the above estimate is considered a worst-case estimate.
3.2. Ionospheric TEC Effects on Signals During Solar Storms
 Global ionosphere maps of TEC are extracted from data acquired from a GPS global monitoring network, a continuously operating world-wide service [Mannucci et al., 1998; Komjathy et al., 2005]. For this service, dual-frequency signals from GPS satellites are received by more than 150 receivers distributed around the globe, each outputting data points every ∼30 s. The slant TEC along the line-of-sight between any satellite-station pair is calculated using an “infinitesimal” thin sheet model of the ionosphere. The TEC estimates are obtained using vertical projection at intersecting line-of-sight points with the shell model and taking into account estimated biases. Differential maps of ionosphere TEC are generated by computing percent change of storm time TEC relative to quiet time TEC. The relative precision of TEC using GPS is about 0.01 TECU2 with an absolute accuracy of 1–3 TECU for vertical measurements [Mannucci et al., 1998]. Mapped vertical TEC using the Global Ionospheric Model Calibration software (GIMCAL) or similar techniques is typically accurate to 5 TECU for most regions [Komjathy et al., 2002a, 2002b; Ho et al., 1998]. From these data, TEC numbers can be estimated for specific applications, such as for flight navigation teams to allow removal or calibration of the ionosphere from tracking data for individual tracks prior to performing trajectory solutions. It is emphasized that more accurate models for TEC and global TEC data are available from the International GNSS Service (IGS) for approximately 400 globally distributed sites [Hernández-Pajares et al., 2009]. By making use of multilayer models such as UPC's TOMION model, multishell GIM [Mannucci et al., 2004] errors in mapping vertical to slant range TEC can be considerably reduced. The use of such models for more detailed analyses is a focus of future work.
 The CMEs associated with the flares of 28–29 October 2003 had propagated to the Earth starting at speeds exceeding 2000 km/sec near the Sun (courtesy of the SOHO/LASCO CME Catalog, NASA GSFC), resulting in a major geomagnetic storm. Partial penetration of interplanetary electric field is believed to have occurred and the associated uplift of the dayside ionosphere is referred to as the “dayside superfountain effect” [Tsurutani et al., 2004, 2008; Mannucci et al., 2005; Huang et al., 2005]. Daytime TEC was estimated along a signal path above the CHAMP satellite (at altitude about 400 km) for several consecutive orbits during the 30 October 2003 event [Mannucci et al., 2005], in which vertical TEC (VTEC) of up to 330 TECU was estimated at magnetic latitudes ∼30°.
 The background daytime ionosphere level is typically ∼100 TECU at solar maximum (middle to low latitudes) whereas enhancements during storms have ranged from about δTEC ∼ 5 TECU over natural TEC variability timescales of minutes to δTEC ∼ 330 TECU over a couple hours.
 For the largest observed increase of 330 TECU, we obtain an estimate of phase fluctuation of 174 cycles or a frequency fluctuation of 0.65 Hz over a 60-s count time using the formulation in section 2 due to the ionosphere. Thus a 40 mm/sec velocity error could be incurred in two-way tracking data acquired during such an event. Note that the peak δTEC was estimated from VTEC measurements for the 30 October 2003 superstorm. For a low-elevation angle signal link, we need to account for a thicker ionospheric layer along the signal path and use slant TEC. Thus, the incurred error could be even higher. We will address these considerations later in section 6.
4.1. Selection Strategy of Tracking Data for Inspection
 Both the direct effect of the presence of the ICME in the near-Earth environment spanning the signal link and the induced disturbance of it or its associated solar flare on the ionospheric portion of the link are to be assessed. The relative level of Doppler excursions for these two contributors can then be compared. In order to assess estimates and bounds of Doppler excursions based on changes in Ne or TEC (section 3), we decided to examine and compare tracking residuals acquired during the strongest solar storms that occurred in recent years.
 We analyzed selected periods of tracking data acquired during the periods of most intense geomagnetic storm activity in the Earth's environment during Solar Cycle 23 associated with ICMEs [Gopalswamy et al., 2007]. These periods were chosen on the basis of the Disturbance Storm Time (Dst) index, a measure of the ability of an ICME to induce magnetospheric currents and generate geomagnetic storms [Sugiura, 1964; Mayaud, 1980]. In order for an ICME to be geo-effective, it must arrive at the Earth with a southward magnetic field component (negative Dst number) [Gonzalez et al., 1994]. A total of 378 halo CME events were observed by the SOHO/LASCO imaging system from 1996 to 2005 [Gopalswamy et al., 2007]. (A halo CME event observed in white light coronal images is usually a good indicator that a CME is Earthward directed.) Of these 378 events, 59 were identified that had Dst < −131 nT, and had originated within 45 deg of the solar disk center. Attempts were made to acquire tracking data from several near-Earth S band missions coinciding with the strongest of these events, however, the cost of retrieving the tracking data for most missions was found to be prohibitive. The comparison of theoretical predictions using independent estimates of TEC and Ne with tracking data was feasible only for the Genesis mission, since the tracking data and required expertise were available at the Jet Propulsion Laboratory in Pasadena, California.
 The Genesis spacecraft was launched on 8 August 2001. The objectives of the mission were to sample solar wind particles (3 December 2001 through 1 April 2004), and to return them to Earth which occurred on 8 September 2004 thus ending the mission. Details of mission design are discussed by Lo et al. . For the purpose of recovering and examining Genesis Doppler tracking residuals, we identified the most intense geomagnetic storms that occurred during the active mission period and selected the three events with the largest negative Dst indices. Table 1 summarizes the periods of the recovered Genesis tracking residuals during which multiple ICME and solar flare events occurred. The data of Table 1 overlap with the strongest geomagnetic disturbances that occurred between 2000 and 2005. These events with the largest disturbance geomagnetic indices (Dst) and associated ICMEs originating within 45 deg of the solar disk center were selected out of a total of 378 total overall events cataloged during 1996 through 2005 [Gopalswamy et al., 2007].
Table 1. Recovered Genesis Tracking Data Periods
Start Time (UTC)
End Time (UTC)
1 Nov 2001
8 Nov 2001
28 Oct 2003
1 Nov 2003
18 Nov 2003
22 Nov 2003
4.2. Genesis Orbit Determination and Generation of Doppler Residuals
Figure 1 displays the Genesis trajectory during the entire mission period in the Earth-Sun rotating frame along with annotation of mission phase with durations. The numbered points on the trajectory mostly coincide with planned trajectory correction maneuvers and provide a visual progression of the spacecraft's progress. What is important here is that the tracking pass data discussed in this paper were acquired while Genesis was in or near its halo orbit about L1, about 0.01 AU on the sunward side from the Earth. Thus, all of these Genesis tracking passes have signal links that are generally in the sunward direction thus intersecting the dayside ionosphere.
 Genesis was a spin stabilized spacecraft with a nominal spin rate of 1.6 rpm. For this study, the spin signatures (due to the offset of the antenna from the center of spin) in the S band Doppler data were removed prior to the estimation process. A standard deviation of 0.3 mm/sec for the Doppler measurements compressed to a 60 s count time is assumed in the Orbit Determination (OD) processing. In order to keep the spacecraft Sun-pointed, the Genesis project also performed small attitude correction maneuvers once a day (to correct for the ∼1° turn each day due to Earth orbital motion). These small maneuvers exerted 3 to 4 mm/sec of ΔV to the spacecraft path, which were also estimated during the OD process. In addition to the daily precession maneuvers, Genesis executed Station Keeping Maneuvers (SKMs) roughly once every two months. The effects of the SKMs were all accurately estimated during the OD process. However, no SKMs occurred within the data periods considered here.
 Other than the spacecraft states, only the solar radiation pressure was estimated using a simple bias model. Time varying stochastic acceleration modeling was not used so that the resulting Doppler residuals (to be analyzed for solar effects) were not contaminated by short-period time varying signatures induced by the OD process itself.
 The Doppler residuals that were generated for this study constitute the observations for which solar storm induced fluctuations were examined and compared with model estimates based on other data such as GPS and spacecraft particle density detector data. These results are discussed in the next section.
5.1. ICME Effects Observed in November 2001 Genesis Tracking Data
Figure 2a displays Genesis Doppler residuals acquired for a selected period (5–6 November 2001 Goldstone downlink pass) near the time of a magnetic storm (seen in SYM-H index). This is one of the three periods of the strongest geomagnetic activity listed in Table 1. Geomagnetic activity during geomagnetic storms is described by Dst index, which is a quantitative measure of disturbances in the Earth's magnetic field derived from measurements at several middle-latitude geomagnetic observatories [Sugiura and Hendricks, 1967]. The Dst index (1-h resolution) is derived and published through the IAGA on a continuous basis. Another index, SYM-H, is essentially the same as Dst, but is derived from a different set of geomagnetic observatories and provided every 1 min (see more in work by Mayaud [1980, and references therein]). Hereafter in this paper we use the SYM-H index as a high-resolution (1 min) geomagnetic activity measure instead of an hourly Dst index (courtesy of the World Data Center at Kyoto University). Here the red curve in Figure 2 denotes the tracking data without ionospheric model calibration. The blue curve denotes the calibrated tracking residuals using the GIMCAL ionospheric model corrections [Komjathy et al., 2002a, 2002b]. There are two tracking passes for which residual Doppler data are displayed; Goldstone (California) Deep Space Station (DSS) DSS-24 spanning from day ∼5.55 (5 November 13:12 UTC) to day ∼5.95 (5 November 22:48 UTC) and Canberra (Australia) station DSS-34 spanning from day ∼5.95 (5 November 22:48 UTC) to day ∼6.17 (6 November 4:05 UTC). There is a 7 1/2 min data gap between the two tracking passes at day ∼5.95, and one three minute gap in the middle of the Goldstone tracking pass near day 5.75 (5 November 18:00 UTC) (both gaps shown by red arrows in Figure 2b). The general effect of applying the ionospheric correction is to flatten the residuals (see Figure 2a). Note that the features occurring shortly before the end of the data appear to be adjusted upwards somewhat after the calibration, but the general character of the signature remains. It is expected that contributions to the Doppler fluctuations are either due to TEC variations over most of the signal link above the ionosphere as the ICME propagates through the near-Earth environment, or due to TEC variations in the ionosphere disturbed by the solar storm. It is also expected that the GIMCAL mapping to the signal path has errors. The calibration technique also may not fully remove the ionospheric contribution during strong disturbances due to ionospheric mismodeling or deficiencies in the employed calibration scheme such as insufficient time resolution. Figure 2d displays SYM-H index for this time period. Note that in Figure 2a large variations in Doppler residuals and a deviation between the calibrated and uncalibrated data start around the onset of this major storm around day 6.1 (6 November 2:24 UTC). For reference, the peak of the geomagnetic storm occurs at 6:00 UTC on 6 November (6.25 days on the time axis in Figure 2a) well after the end of the tracking data. It is thus expected that such variations could be larger beyond the available data shown in Figure 2a. There were no available Genesis data until the next Canberra tracking pass starting at day 6.955 (6 November 22:55 UTC) through day 7.08 (7 November 1:55 UTC) by which time the Doppler data were nominal and representative of non-storm conditions.
Figure 2b displays relative column density extracted from the phase measurements during this period. Based on the red curve in Figure 2b which includes the contribution of the unmodeled ionosphere, we see excursions on order of 60 TECU. Based on a visual inspection of the calibrated data (blue curve) where an ionosphere model was removed, we see an overall excursion of about 20 TECU over the pass with some smaller excursions or wiggles on order of 5 to 10 TECU over 1 to 5 h periodicities. There were two short outage periods (indicated by red arrows in Figure 2b) where there may be some small amount of uncertainty in connecting the curves but is not expected to be significant given that these gaps are of very small time durations.
Figure 2c displays proton number density data directly measured by the Genesis ion counter for this period. The data were acquired while Genesis was on the sunward side of the Earth prior to L1 halo orbit insertion (LOI) near point 6 on the trajectory in Figure 1. Figure 3 displays the proton data acquired from Genesis over a longer several day period that included the period of the 5–6 November 2001 tracking pass (indicated by arrows). From Figure 3, one can discern the relative behavior of the longer quiet period relative to the active period starting about 5.6 day (5 November 14:24 UTC). One can clearly see a plasma structure with sharp density increase to ∼70 cm−3 passing by the spacecraft around 5.65 day (5 November 15:36 UTC) following by another increase around day 5.8 (5 November 19:12 UTC) and relatively small increase just after 6.0 day (6 November 00:00 UTC). We will discuss how the timing of these structures corresponds to variations in Doppler residuals (Figure 2a) and disturbances in TEC corrected by calibration (Figure 2b blue curve). It should also be noted that there were numerous dropouts in the proton instrument data as suggested by a close inspection of Figures 2c and 3.
 Solar wind parameters at the Earth orbit taken from the OMNI database show a passage of an interplanetary structure at 18:00 UT (∼5.75 day) which triggered a moderate magnetic storm (SYM-H exceeds -100 nT) seen in Figure 2d. This agrees with the Genesis plasma measurements. Assuming solar wind speed of 400 km/sec (OMNI data) it took ∼1 hr (less than 0.04 day) for the plasma structure registered at Genesis to reach the Earth and cause a geomagnetic disturbance. The onset of a major geomagnetic storm occurred with arrival of the CME cloud around ∼6.1 day (6 November 02:24 UTC). Measurements from the ACE spacecraft which is also located near L1 confirm the passage of an interplanetary structure suggested in the Genesis data. The particle detector data from Genesis although of better quality than that from other spacecraft (ACE, WIND and OMNI), did have cases of numerous dropouts. Unfortunately, the OMNI database (ACE data) and WIND measurements database have a data gap during the main phase of this storm. The Genesis data (Figure 2c) does not show an interplanetary density increase comparable to the increase on the previous day.
 We will discuss three large excursions in ionospheric TEC (indicated by the event numbers labeled on Figure 2b). Two large peaks in the interplanetary proton density in Figure 2c occurred between days 5.6 and 5.8 (5 November 14:24 to 19:12 UTC) and before the magnetic storm. These could cause the small variations ∼0.01 Hz in Doppler residuals (Figure 2a). A fluctuation in δne of ∼40 cm–3 was used to estimate the interplanetary contribution and was inferred directly from the Genesis data of Figure 2c. If we assume this value is representative of the density variations sustained over the signal link between Genesis and the ground station, then the resulting estimate of column density fluctuation of 6 TECU is consistent with the magnitude of the 5 to 10 TECU peak-to-peak periodicities in the ionosphere model-calibrated column density (blue curve) shown in Figure 2b for the events labeled 1 and 2. This estimate assumed a link distance L of about 3 Earth-Moon distances (point 6 in Figure 1) as the ICME propagated from Genesis to the Earth and also assumed incoming ICME velocities ∼400 km/sec (Coordinated Data Analysis Website (CDAWeb), http://cdaweb.gsfc.nasa.gov/istp_public/) over ∼1 h time periods, the time for which the disturbances would propagate over the signal path.
 It is unlikely that a geomagnetic disturbance caused a large correction (red to blue curves) in TEC near day 5.75 (5 November 18:00 UTC) (shown near the first red arrow in Figure 2b and corresponding to event 2). The SYM-H value was positive at this time (see Figure 2d). Possible explanations for the large excursion between the blue and red curves for the event 2 include local time (LT) and elevation angle effects. The elevation angle was near its maximum value for the Goldstone tracking pass using station DSS 24. Thus larger corrections are expected at daytime due to the larger ionospheric TEC and ionosphere thickness values as compared to nighttime values.
 The variation in Doppler shown near the end of the data (near day 6.14 or 6 November 03:22 UTC) in Figure 2a of ∼0.03 Hz (blue curve after the ionosphere model correction) is consistent with the predicted 0.06 Hz magnitude for the interplanetary component (using section 2 formulation with 40 cm−3 particle number density and 60-s count time). However, Genesis did not register large changes in solar wind density around that time. The corresponding short-term few-hour time scale wiggles in column density on the blue curve in Figure 2b (∼5 TECU) appear to be consistent with interplanetary charged particle variations along the signal path above the ionosphere.
 On the other hand, one would expect strong ionospheric response to this large geomagnetic storm (SYM-H < −300 nT) which also might cause the disturbances in Doppler residuals near day 6.14 during the DSS-34 pass. The large TEC correction (event 3 in Figure 2b) indicates a highly disturbed ionosphere. The effect of the ionospheric GIMCAL correction in the TEC estimates (difference between the red and blue curves in Figure 2b) of up to 90 TECU dominates over the smaller level of variations seen in the blue curve in Figure 2b. The higher level of TEC variations due to the ionospheric model correction such as the 70 TECU over 0.3 days at the end of the DSS-34 tracking pass suggests that most of these variations during the storm around ∼ day 6.1 (6 November 02:24 UTC) are due to ionospheric effects.
5.2. ICME Effects Observed in October 2003 Genesis Tracking Data for the Halloween Event
 The Halloween 2003 event was among the strongest earthward directed ICMEs that occurred during the Genesis mission. Figure 4 displays Genesis two-way Doppler residuals for the Madrid and Goldstone tracking passes on 30 October 2003 with and without ionosphere calibration applied. Figure 5 displays the corresponding range rate residuals for the Goldstone tracking pass. The Madrid data (see Figure 4) (day 30.3 to 30.5) (30 October 07:12 to 12:00 UTC) before the major ICME event (30.7 to 30.9) (30 October 16:48 to 21:36 UTC) appear fairly stable and show very little variation but a biased trend is apparent for the uncalibrated ionosphere case likely due to changes in the ionospheric path as the elevation angle changes. The later Goldstone data at the right in Figure 4 show large fluctuation features with excursions as large as ∼0.3 Hz (without ionosphere calibration) and about 0.2 Hz (with ionosphere calibration). The ionosphere model contribution appears to run about 0.15 to 0.2 Hz and shows long temporal variations which flatten out by day 30.88 (30 October 21:07 UTC). A large dip of 0.2 Hz in the calibrated data at the very end of the interval is seen on the blue curve in Figure 4. It can be attributed to non-ionosphere effects of the ICME passage along the signal link as well as possible deficiencies in modeling ionospheric storm effects.
 The Halloween events of 29–30 October 2003 were complex solar and interplanetary events. Two X-class solar flares occurred on 28 October at 09:51 UTC and 29 October at 20:37 UTC, consecutively [Yurchyshyn et al., 2005]. A duration of the enhanced level of TEC of ∼3 h was reported for the 28 October event [Tsurutani et al., 2005]. Thus we do not consider that it affected the ionosphere state during the following day. The 29 October flare caused a peak daytime TEC increase of ∼5 TECU above a subsolar point background value of 90 TECU [Tsurutani et al., 2005] which is less than our estimates of a TEC disturbance. Here we discuss measurements made during the Goldstone tracking pass, where the site location is about 243° in longitude. The flare event corresponds to ∼04:25 LT at the site and the Sun was below the horizon. Thus, we do not consider that flare effects were a significant contributor to the fluctuations seen in our measurements.
 The large feature seen in Figures 4 and 5 is shown for before (red curves) and after (blue curves) ionospheric model calibration. This event occurs right about the time of expected ICME arrival on 30 October 22:00 UTC (or day 30.92) (based on measured peak Dst) [Gopalswamy et al., 2007]. The effect of the GIMCAL correction flattens the residuals and reduces the level of the excursion from 0.3 Hz to 0.2 Hz during the ICME, similar to that seen for November 2001 discussed in section 5.1. The reduction of this excursion suggests that a sizable fraction of the contribution is due to ionosphere disturbance. The corresponding change in velocity is about 20 mm/sec over 5 h (Figure 5).
 During this data acquisition period in October 2003, Genesis was located in its halo orbit about L1 during its science phase (see Figure 1). The excursion of the Doppler is about 0.3 Hz over a 0.2 h period and would have continued past the end of the tracking data period displayed in Figure 4. The ∼0.1 Hz DC bias shown in Figure 4 is the result of the large excursion not being modeled out influencing the original solution. Such a bias is not present or at least not discernible in the November 2001 data of Figure 2a where a much smaller magnitude variation occurred.
 In order to test whether any information was lost in the original ionospheric calibrations, the GIMCAL data were processed using a higher temporal resolution polynomial over the Goldstone tracking pass. It was found that there is little difference between the different GIMCAL ionosphere calibrations that were employed. This suggests that the ionosphere accounts for a long period trend of about 0.07 Hz which removes much of the overall variation up to day 30.88 (30 October 21:15 UTC) leaving some smaller wiggles and the large 0.2 Hz excursion at the end of the data. The ∼0.2 Hz excursion could be due to ICME disturbances on the signal link above the ionosphere, but given the significant nature of this event, the thin-layer GIMCAL model used may also be susceptible to larger errors.
 Here we focus on the main phase of the magnetic storm after ∼30.8 day (30 October 19:12 UTC) (see Figure 6c). For an ICME-type storm it is attributed to a passage of magnetic cloud which is characterized by low and nearly constant plasma density. The major drop in the Doppler residuals occurring around that time cannot be explained by interplanetary factors.
 During the 30 October 2003 Halloween event, Mannucci et al.  measured about 900% increase in electron content above the CHAMP satellite altitude (400 km) at midlatitudes (where Deep Space Network (DSN) tracking stations reside) using GPS measurements of the ionosphere. Here the geomagnetic storm-time phenomenon of prompt E-field penetration was cited as a possible factor causing the increased electron content. This resulted in day-side ionosphere uplift (daytime super-fountain effect in the ionosphere) [Mannucci et al., 2005]. By far the largest ground TEC trace was measured between 21:43 and 22:04 UTC as the CHAMP satellite passed over Goldstone with peak TEC at around 30° mid latitudes [see Mannucci et al., 2005, Figure 3]. The Genesis to Goldstone link did transit near this signal path, as Genesis was located in the western sky as seen from California.
 Assuming a change of 330 TECU (see section 3.2) measured for the ionosphere during the Halloween event [Mannucci et al., 2005], we obtain an estimate of phase fluctuation of 174 cycles or a frequency fluctuation of 0.65 Hz at S band with a 60-s count time using formulation in section 2. Thus, a 40 mm/sec velocity error could be produced in two-way tracking data acquired from such a link. The maximum observed excursion of ∼0.3 Hz (Figure 6a, which is the Goldstone data in Figure 4 repeated) is a reasonable fraction of this bound, suggesting that the fluctuations are mostly of ionospheric origin. Electron density fluctuations of many times greater would be needed to account for the magnitude of the non-ionospheric Doppler contribution (∼0.2 Hz) seen just before the end of the data shown in Figure 6a. However, assessing this would require in situ particle detector data. An attempt was made to retrieve ion counter data from the Genesis data archive for the purpose of correlating it with the Doppler fluctuations. However, it was found that the data for several days around the Halloween event period were not available due to an instrument safing event attributed to the ICME itself (R. Wiens, Los Alamos National Laboratory, private communication, 18 September 2007).
 The polarity of the associated magnetic field (Bz) is the major factor contributing to geoeffectiveness. If Bz is in the opposite sense as Earth's magnetic field (southward), the magnetic fields will more easily reconnect causing magnetic and ionospheric storms as was the case for the 30 October 2003 Halloween event [Tsurutani et al., 2008]. The ACE spacecraft did not measure a significant increase in proton number density during this event and we do not estimate an interplanetary contribution to the Doppler residuals here. However, the satellite measured high velocities above 1500 km/sec and a large southward Bz component just before the event [Mannucci et al., 2005; Skoug et al., 2004]. Figure 6b displays a profile of electron column density extracted from the radio phase data of Figure 6a for both cases with and without ionosphere model calibration (an arbitrary value of 0 TECU was used to anchor the two curves at the beginning near day 30.72 (30 October 17:17 UTC), thus the differences between the two curves for any later point in time should be representative of any variation relative to the anchor point). Figure 6c displays SYM-H index during and around this tracking pass period to illustrate the progress of the magnetic storm. Note that the ionosphere model contribution is ∼700 TECU (difference between red and blue curves at last data point) which is a slant TEC or estimate of TEC along the line of sight. Mannucci et al.  have shown that verticalized TEC of ∼330 TECU was observed above the CHAMP satellite (up above 400 km orbital altitude). Conversion from slant to a vertical TEC estimate in our case with the elevation angle of 22° at 21:00 UTC (day 20.9) could decrease our ∼700 TECU estimate by a factor of 2.6 resulting in ∼270 TECU which is in reasonable agreement with the CHAMP results. The absence of particle data on Genesis makes it difficult to disentangle the ionosphere and non-ionosphere contributions, but it appears that the ionospheric disturbance is the main contributor to the observed tracking data excursions for this solar storm.
 Interplanetary parameters for the 29–30 October 2003 events were extensively studied by Skoug et al. . ACE magnetic field and plasma measurements (their Figure 3) indicate a magnetic cloud between two interplanetary shocks at ∼29.2 (29 October 04:48 UTC) and 30.6 days (30 October 14:24 UTC), immediately followed by another magnetic cloud. The latter caused a magnetic storm which commenced around 30.78 day (30 October 18:43 UTC) and reached the minimum value of Dst = −390 nT at ∼30.98 day (30 October 23:31 UTC) (see details in work by Tsurutani et al. [2005, and references therein and Figure 1]). ACE observed passage of several structures (shocks, magnetic clouds) during these two days. Southward turnings of the interplanetary magnetic field which caused geomagnetic activity were alternating with occasional northward turnings. These changes in magnetic field orientation together with interplanetary shock impacts resulted in variations in SYM-H index (Figure 6c). The density measurements, though uncertain at such highly disturbed conditions, showed a cold low density (<10 cm−3) plasma interval from day ∼29.2 (20 October 04:48 UTC) to 31.0 (31 October 00:00 UTC) [Skoug et al., 2004] which is unlikely to cause large variations in Doppler residuals.
5.3. ICME Effects Observed in November 2003 Genesis Tracking Data
Figure 7a displays Genesis Doppler residuals for 19–20 November 2003, along with annotation that identifies DSN tracking passes for individual 34-m diameter subnet antennas (DSS-34 near Canberra, Australia, DSS-54 near Madrid Spain and DSS-24 in Goldstone California). Here, another significant earthward directed ICME produced a magnetic super-storm with a large peak Dst value of −422 nT on 20 November 2003 at 21:00 UTC (or day 20.875). ICME arrival and onset of the storm occurred at about 08:00 UTC or day 20.33 [Yizengaw et al., 2006]. A relatively clean background trend from 19.9 days (19 November 21:36 UTC) up to near 20.35 (20 November 08:24 UTC) days is evident in Figure 7a during most of the Canberra DSS-34 tracking pass followed by many large Doppler variations near the vicinity of the main phase of the magnetic storm (seen in the negative SYM-H index in Figure 7d) during the Madrid DSS-54 and Goldstone DSS-24 tracking passes. The ionosphere calibration flattens the residuals somewhat but leaves the general signatures relatively intact suggesting that the fluctuations may be due to TEC variations above the ionosphere or mapping errors in the ionosphere calibration similar to what was hypothesized for the November 2001 (section 5.1) and October 2003 (section 5.2) events. The largest fluctuation feature of ∼0.06 Hz seen in the November 2003 data (Figure 7a) lies well below the 0.2 Hz magnitude fluctuation observed during the October 2003 Halloween event, but is consistent with expected levels.
Figure 7b displays the relative column density extracted from the data in Figure 7a for each of the three tracking passes. It is emphasized here that the values assigned to the start of each tracking pass segment is arbitrary and thus set equal to zero at the start of each tracking pass, as it is not possible to discern the absolute behavior in the gaps between tracking passes. What is important is the relative behavior of the column density within each of the continuous arcs of data. Note that the ionosphere model tends to flatten the TEC signatures, but there are cyclic-like variations of up to ∼30 TECU over periods of ∼hours occurring during the last two tracking passes that coincide with ICME activity. The uncalibrated (red) column density curves show some similarity and large magnitude trending relative to the calibrated model (blue) column density curves. This level of TEC variation is consistent with midlatitude daytime TEC variations up to 40 TECU observed with GPS above midlatitude stations [Yizengaw et al., 2006].
Figure 7c displays proton number density as measured by the Genesis on-board ion-counter coinciding with the time frame in Figure 7a. The low values of 5–10 cm−3 at the start (see Figure 7c) prior to the ICME are consistent with expected quiet background levels. For the L1 to Earth link, the column density corresponding to this electron number density is ∼1.2 TECU (neglecting ionosphere). The increased number of protons during the ICME arrival period is evident in Figure 7c starting near day 20.35 (20 November 08:24 UTC), and coincides with a period of large Doppler fluctuations in Figure 7a. Peak values of ion number density approaching 80 cm−3 are within the bounds expected for such events. We assume that proton number density is close to electron number density (plasma charge neutrality). The Doppler data displayed in Figure 7a is well behaved and quiet at the start of the displayed period where the proton counts in Figure 7c are minimal. During the later active period, there are large fluctuations with some gaps in the Doppler record which coincide with significantly increased proton counts. If we assume that the sustained long period of increased charged particle density at and above 40 cm−3 around day 20.35 (20 November 08:24 UTC) to day 20.45 (20 November 10:48 UTC) in Figure 7c is representative of the density fluctuation over the interplanetary signal link, then the inferred TEC variation is consistent with the column density excursions (extracted from the tracking phase data) near day 20.5 to 20.6 (20 November 12:00 to 14:24 UTC) in the blue curve of Figure 7b. Thus, this suggests that a portion of the excursion in Doppler could be attributed to the ICME activity along the signal path above the ionosphere. The variations seen in the Goldstone DSS-24 pass between day 20.7 to 20.95 (20 November 16:48 to 22:48 UTC) can be attributed to a combination of storm effects on the ionosphere not fully removed by the model (high negative SYM-H values shown in Figure 7d) and interplanetary component values (high charged-particle number density values over an extended time shown in Figure 7c). Figure 8 displays a longer time record of proton number density beyond the data span of Figure 7 to be used for visualization of the sustained background levels around the period of this event.
6. Prediction of CME Arrivals at Earth and Mitigation of Signal Link Effects
 Making use of available solar event forecasting or space weather services is recommended to produce alerts for any such events that may impact spacecraft operations during critical navigation data acquisition periods. Woo  discusses various methods to make use of daily observations from missions such as SOHO to diagnose, monitor and forecast adverse space weather conditions on deep space communication. The higher the link frequency, the less susceptible the link is to charged-particle density induced effects. One suggested recommendation to minimize such effects would be to use higher frequency links such as Ka band or even optical communications.
 The utility of Ka band would be further enhanced by maintaining dual-frequency links (S and Ka) which would provide for a calibration of the charged-particle media [Iess et al., 1999]. Even though the vehicle itself may not be provided with a ground-navigation-independent means of coping with charged-media effects, the dual-frequency calibration would virtually eliminate (or significantly reduce) a returning trans-lunar vehicle's susceptibility to such effects when ground tracking was available.
 In order to identify signatures in S band Doppler that are caused by ICME events and remove the associated data from trajectory solutions, we can partially rely on ionospheric maps such as provided by GIMCAL or UPC's TOMION services. However, some fraction of ICME effects may be attributed to charged particles above the ionosphere as the section 5 analysis suggests. It is therefore recommended that studies be conducted that explore the use of a network of satellites to monitor the charged-particle environment to determine periods of contaminated data so that they can be removed from consideration in the trajectory solutions. For trans-lunar vehicles, it may be beneficial to include instrumentation on board such as ion counters to provide ancillary data that could be monitored and used to correlate with such events. Such data together with data from other spacecraft systems could be combined to allow identification of potentially corrupted data periods, or even be combined to calibrate out such trends. Future work would entail recovering electron number density data from the various satellites in the near-Earth vicinity during periods of CME arrival such as was done here with the Genesis tracking data periods and correlate them using finer time resolution. An optical navigation system on-board the spacecraft could be used to negate such concerns.
 Since ionospheric storm-time effects on signal links are very important and can be significant, it is also recommended that the tracking stations be located at latitudes where the contributions due to ionospheric disturbances are minimal, and perhaps the signal paths could be timed such that expected occurrence of disturbances could be minimized (spatial diversity of multiple antennas on the ground).
 We have provided theoretical background and estimation for quantification of ICME-associated interplanetary electron number density changes along the signal path based on measured spacecraft-to-ground signal phase fluctuations. We have also quantified the level of such effects on S band signal phase with ionospheric electron density disturbances over the signal path associated with magnetic storms. We have also recovered, reprocessed and presented Genesis S band tracking data acquired during periods of large ICME arrivals at Earth and we have compared the observed degraded Doppler data with the predictions based on expected electron column density levels. We have shown from both theory and measurements that ICMEs can induce significant trends in received spacecraft Doppler that can potentially mask Earth gravity field effects in navigation trajectory solutions at the nominal S band link frequency used for near-Earth communications and navigation. Based on the analysis performed here, it appears that ionospheric disturbances of up to several hundred TECU produced by extreme solar storms caused by ICMEs are the most significant contributors to tracking phase excursions on the near-Earth S band signal links. These disturbances can induce trends of ∼0.3 Hz or more over ∼hour periods, which can corrupt OD solutions, that could be critical such as during Earth reentry targeting. Very large electron number density variations of ∼100 cm–3 in the near-Earth environment caused by incoming solar storm events can induce significant long period velocity excursions in space tracking Doppler reaching values of ∼50 mm/sec, lying well above the noise level of the data. Excursions in range rate as high as 20 mm/sec have been observed in the Genesis tracking data and have been attributed to such events. Finally, we have suggested a few ways that such effects can be identified, alleviated or eliminated.
 We would like to thank the NASA Constellation Project for supporting the initial phase of this work, and the Jet Propulsion Laboratory (JPL) Division 300 Raise the Bar (RTB) Program (Stephen Lichten, Peter Theisinger and Adriana Wall) for supporting an intermediate phase of this work. We would like to thank Charles Naudet of the Advanced Engineering Program at JPL for supporting the concluding phase of this work. We would like to thank Anthony Mannucci of JPL for his valuable comments on an early draft; Tom Runge and Attila Komjathy of JPL for assistance in providing TSAC ionospheric calibrations (GIMCAL) used to adjust Genesis tracking data; and Roger Wiens and the Los Alamos Genesis Science Team for providing the Genesis particle detector data used in the correlative analysis discussed in this article. The research described in this paper was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. Government sponsorship acknowledged.