Radio Science

Determinations of plasmasphere electron content from a latitudinal chain of GPS stations



[1] An extended application of a regional method (SCORPION) for calibration of GPS total electron content (TEC) measurements and plasmasphere electron content (PEC) determinations is described and demonstrated for a particular chain of GPS stations, in order to resolve the ambiguity between an omnipresent PEC contribution and the derived receiver system bias. Based on these calibrations and PEC determinations, a latitudinal profile of vertical TEC is presented for North American stations ranging from Alaska to central Mexico, with separate profiles for both the ionosphere (nominally to 1000 km altitude) and the plasmasphere. Determination of the local vertical PEC remedies a latitudinal TEC “sawtooth” feature arising from the standard conversion of slant TEC to equivalent vertical TEC at each station. Comparisons to other TEC methods and measurements, including SCORE, Jason-1, and IONEX, are also presented. SCORE biases derived for simulated data confirm a latitudinal trend in its bias errors that is also evident in comparisons to SCORPION and Jason-1. IONEX vertical TEC values are generally greater than the SCORPION vertical TEC values, although the associated comparisons of IONEX to SCORE results indicate that this is primarily a consequence of the conversion from slant TEC to equivalent vertical TEC without separating the ionosphere and plasmasphere components.

1. Introduction

[2] In recent years, techniques have been developed to allow determinations of plasmasphere electron content (PEC) from ground-based GPS total electron content (TEC) measurements, for single stations [e.g., Mazzella et al., 2002, 2007; Anghel et al., 2009; Carrano et al., 2009], paired stations [e.g., Lunt et al., 1999d], and networks of stations [e.g., Schunk et al., 2004]. Distinct and supplementary determinations from satellite-based GPS receivers have also been developed [e.g., Yizengaw et al., 2005, 2008; Spencer and Mitchell, 2011]. An additional objective of the ground-based single station developments was the improvement of bias calibrations of GPS TEC measurements by accounting for the complicating influences of the plasmasphere on determinations of ionosphere electron content (IEC) [Lunt et al., 1999b; Mazzella, 2009].

[3] This study presents a determination of PEC using the SCORPION method [Mazzella et al., 2002, 2007], including a process to resolve the ambiguity between an omnipresent (regional) PEC contribution and the receiver system bias by employing a latitudinal chain of receiver stations [Mazzella et al., 2007]. For comparison with Bishop et al. [2009], data from a chain of stations along the west coast of North America were used, for the same date (8 April 2007). The study described here also includes more northerly Alaskan data, where minimal PEC effects are expected for overhead measurements. The date and geophysical conditions of this study are also similar to those for a model study of plasmasphere effects and United States east coast measurements presented by Mazzella [2009].

[4] Sixteen stations were selected for processing (see Figure 1 and Table 1), to provide latitudinal overlap between stations while spanning a region from minimal plasmasphere content (Fairbanks, AK) to a region with significant plasmasphere content (Colima, Mexico). The observational regions (ovals in Figure 1) for each station correspond to elevations above 35°.

Figure 1.

The 16 stations used for the IEC and PEC determinations, plus the Kodiak Island (KOD2) station referenced for the determination of the Jason-1 bias. The ovals around each station display their observational regions at 350 km altitude for a threshold elevation of 35°, while the smooth curved line along the coast indicates the ground track of the Jason-1 satellite.

Table 1. The Identifications and Locations of the 16 Stations Used for the IEC and PEC Determinationsa
Site NameSite IdentifierGeographic Latitude (deg)Geographic Longitude (deg, +E)Magnetic Latitude (deg)Magnetic Longitude (deg, +E)
  • a

    Except for the data from Holberg, Canada, which were obtained through the IGS network, all of the GPS data for these stations were obtained through the CORS network.

Cape HinchinbrookCHI460.238−146.64760.504−93.766
Level IslandLEV256.466−133.09259.724−79.009
Neah BayNEAH48.300−124.62053.499−66.368
Fort StevensFTS146.210−123.96051.536−64.845
Point LomaPLO532.670−117.24039.087−53.405
La PazLPAZ24.140−110.32031.418−44.172

[5] As in the previous study by Mazzella [2009], the boundary between the ionosphere and plasmasphere is considered to be at an altitude of 1000 km, for all latitudes. This corresponds approximately to the altitude of the Transit (or Navy Ionospheric Monitoring System (NIMS)) satellites, but is below the altitude (1336 km) of the TOPEX, Jason-1, and Jason-2 satellites and above the altitude (approximately 800 km) of the COSMIC satellites. These low Earth-orbit (LEO) satellites all allow (different) operational measurements of IEC, with a corresponding PEC derived as the difference between GPS TEC and LEO IEC measurements [Ciraolo and Spalla, 1997; Lunt et al., 1999c]. This operational altitude boundary between the ionosphere and plasmasphere is distinct from a boundary based on geophysical parameters, such as the O+/H+ transition altitude [e.g., Lunt et al., 1999a].

2. GPS TEC Determinations

[6] The calibrations of the biased GPS TEC measurements were performed using the SCORPION method [Mazzella et al., 2007], which also determines the PEC. The SCORPION (SCORE for Plasmasphere and Ionosphere) method is an extension of the earlier “Self-Calibration of Range Error” (SCORE) method [Bishop et al., 1994], which used comparisons of equivalent vertical TEC (Vαi, Vβj) to determine the combined receiver (Br) and satellite (Bα) biases, for the calibration of the GPS TEC measurements. The basis for these determinations is the minimization of the quantity

display math

where Wαiβj is a weight factor (depending upon the elevations and separations in latitude, local time, and measurement time between the data samples αi and βj, for different satellites α and β and respective samples i and j). The distinct equivalent vertical TEC measurements Vαi and Vβj incorporate terms for bias corrections. SCORPION imposed an explicit designation of these equivalent vertical TEC measurements as ionospheric measurements, with the plasmasphere contribution removed, so that the expression defining the equivalent vertical IEC measurements is

display math

Here, f(εαi) is a slant-to-vertical conversion factor for IEC, as a function of elevation (εαi) for the measurement, Mαi is the (biased) slant TEC measurement, and P(εαi, ωαi, tαi) is the (slant) PEC for elevation angle (εαi), azimuthal angle (ωαi), and time (tαi) [cf. Mazzella et al., 2002]. P(εαi, ωαi, tαi) is implemented by a parametric representation of the plasmasphere, with parameter values that are determined by the same minimization process as the biases [Mazzella et al., 2002, 2007].

[7] The determinations of the PEC, IEC, and biases by SCORPION were previously validated using simulations of GPS TEC provided by the Sheffield University Plasmasphere Ionosphere Model (SUPIM) [Bailey and Balan, 1996], for four sites between magnetic latitudes 0° and 45° [Mazzella et al., 2007], in a manner similar to that previously employed by Lunt et al. [1999b]. For low solar flux conditions, these validations demonstrated bias determination errors of 1.5 TEC units or less (1 TEC unit = 1016 electrons/m2 column density), and typical errors of 2 TEC units or less for the equivalent vertical IEC and slant PEC. Experimental corroborations of SCORPION TEC determinations, primarily for the ionosphere, were conducted at Ancon, Peru, against integrated densities from the Jicamarca radar, and at CheCheng, Taiwan, against NIMS and Jason-1 TEC measurements, for various solar flux levels [Mazzella et al., 2007]. The equivalent vertical IEC derived by SCORPION typically agreed with the corresponding radar, NIMS, and Jason-1 TEC values to within about 5 TEC units, for peak equivalent vertical IEC values of about 30 to 120 TEC units. In addition to improvements in the TEC and bias determinations (relative to SCORE), the interpretation of the TEC measurements (by partitioning these into separate ionosphere and plasmasphere contributions) was also enhanced. Further aspects of this partitioning were discussed by Mazzella [2009].

[8] Presuming an appropriate parametric representation of the plasmasphere and a unique minimization for equation (1), a solution ambiguity still remains in the presence of a nonzero minimum slant plasmasphere electron contribution (the “plasmasphere baseline”). This is manifested as a shift in the derived biases, because the sum (Bα + Br) + P(εαi, ωαi, tαi) is unchanged whether this “plasmasphere baseline” constant is assigned to the biases or the plasmasphere. The derived IEC is also unchanged, but the composite TEC (ionosphere plus plasmasphere) is changed based upon this “plasmasphere baseline” assignment. Mazzella et al. [2007] noted that a latitudinal chain would be one technique for resolving this ambiguity in the slant PEC, and that technique is described and demonstrated here.

3. Inter-station Comparisons for Plasmasphere Baseline

[9] Figure 2 displays a SUPIM simulation of GPS TEC results, in a form typically used for SCORPION. The simulated data for the slant PEC are displayed in Figure 2 (top), and equivalent vertical IEC data are displayed in Figure 2 (middle), while the magnetic latitudes of the Ionosphere Penetration Points (IPPs) are displayed in Figure 2 (bottom). The horizontal axis displays the magnetic local time at the IPP, defined to occur at 350 km altitude. The displayed values are restricted to elevations above 35°, to avoid excessive error from the slant-to-vertical conversion factor [Andreasen et al., 1998b].

Figure 2.

SUPIM simulation for GPS TEC for Colima, Mexico, for 8 April 2007, with (bottom) IPP magnetic latitudes (MLAT), (middle) equivalent vertical IEC (EQ VER TEC), and (top) slant PEC (PSPH STEC), all versus magnetic local time (IPP MAG LOCAL TIME) at the IPP.

[10] A notable feature of Figure 2 is that the minimum slant PEC is nonzero, allowing an ambiguity between the derived biases and the plasmasphere baseline. (Note that a minimum derived PEC of zero for a single-station calibration does not preclude an actual nonzero plasmasphere baseline.) Comparison of PEC values to a nearby poleward station could provide a resolution of this situation, provided that the ambiguity has been resolved for the poleward station. Iteration of this pairwise comparison process implies a chain of stations, with the ultimate reference station in the sequence occurring sufficiently poleward so that the plasmasphere baseline is known to be absent or can be demonstrated as absent by some measurement.

[11] The quantity characterizing the PEC for the inter-station comparisons must also be considered with care. The slant PEC itself is unsuitable for this role, because of the extensive spatial distribution of the plasmasphere densities along the line-of-sight and the typical latitudinal separation between the IPP and the corresponding penetration point for the ionosphere-plasmasphere boundary [Mazzella, 2009]. The SCORPION representation for the plasmasphere provides a suitable alternative for these comparisons, using the local vertical PEC at the IPP, even though this quantity is not directly available from the GPS TEC measurements, because of the geometry associated with those observations. For the example depicted in Figure 3, stations at Colima, Mexico, (COL2) and La Paz, Mexico, (LPAZ) view a common IPP (in latitude and local time coordinates) midway in latitude between the stations. The local vertical PEC at the IPP corresponds to the integrated electron content upwards from 1000 km along the vertical line midway between the stations. For appropriately derived plasmasphere representations at the two stations, the diurnal patterns of the local vertical PEC should match, except for the possible baseline ambiguity. This comparison is depicted graphically in Figure 4, using SCORPION calibrations of SUPIM simulations. If the baseline ambiguity has already been resolved for the poleward station of the pair (La Paz), then the correction for the equatorward station (Colima) is determined from the difference of the diurnal local vertical PEC patterns. (For this example, the plasmasphere baseline was determined for La Paz by comparison to SUPIM.) The adjusted slant PEC for Colima compares well with the true slant PEC simulated from SUPIM (Figure 5). Note that the compensatory bias shift has been invoked, so that the derived equivalent vertical IEC values remain unchanged by the plasmasphere baseline adjustment.

Figure 3.

Latitude/altitude depiction of lines-of-sight through a common IPP (at 350 km) for Colima, Mexico (COL2), and La Paz, Mexico (LPAZ), indicating the common observed ionosphere regions and distinct observed plasmasphere regions (above 1000 km), as well as the common vertical plasmasphere region obtainable from the SCORPION plasmasphere representation.

Figure 4.

SUPIM simulations of GPS TEC for La Paz, Mexico (poleward station), and Colima, Mexico (equatorward station), as calibrated by SCORPION, but with a supplementary absolute plasmasphere baseline determination for La Paz only. In contrast to Figure 2, Figure 4 (top) (PSREP VTEC) displays the vertical PEC for the IPPs displayed in Figure 4 (bottom).

Figure 5.

The SCORPION results (red) for the SUPIM GPS TEC simulation (blue) for Colima, Mexico, with the SCORPION plasmasphere baseline derived from the comparison of the Colima and La Paz plasmasphere values displayed in Figure 4.

4. Data Analysis

[12] The GPS data for each station for 8 April 2007 (day 98) and 9 April 2007 (day 99) were obtained from the NOAA Continuously Operating Reference Station (CORS) and International GNSS Service (IGS) archives, and the two days were combined and processed using the GPS Toolkit [Tolman et al., 2004], with phase discontinuity corrections and the calculation of azimuth and elevation values. Incomplete satellite passes at the beginning of the time period were eliminated from processing, and passes commencing after 00:00 UT on 9 April 2007 were also excluded, to provide complete satellite passes spanning a period of approximately 24 h, associated with 8 April 2007. Supplementary processing, following the alignment of dispersive carrier phase to dispersive group delay, was conducted to diminish residual effects of multipath, according to general principles described by Kee and Parkinson [1994] and a particular algorithm described by Andreasen et al. [2002].

[13] GPS TEC calibrations were conducted using SCORPION [Mazzella et al., 2007], individually for each station (rather than utilizing the multiple-station capabilities adapted from SCORE [Andreasen et al., 1998a]). The plasmasphere conditions determined for Neah Bay (NEAH), Washington, prompted a further poleward extension of the chain of stations used by Bishop et al. [2009] to Fairbanks (FAIR), Alaska, to assure attainment of a minimal plasmasphere baseline at the most poleward station in the chain. The situation for Neah Bay was not unexpected, based on previous measurements conducted at nearby Bellevue, Washington, during the development and validation of SCORPION [Mazzella et al., 2007]. A data outage of about an hour was observed for the data from Bakersfield (BKR1), California, so the nearby station at Cambria (P067), California, was used instead. The additional stations of Hermosillo (HER2), Mexico, and Aguascalientes (INEG), Mexico, were included in the equatorward portion of the chain, for greater latitudinal overlap among stations in that region.

[14] Results from these calibrations, prior to any plasmasphere baseline corrections, are displayed in Figures 6a and 6b. A noteworthy feature is the similarity in IEC levels and temporal profiles throughout the high and middle latitude stations. The northern crest of the equatorial anomaly dominates the daytime profile for the southernmost stations, with an associated increase in the plasmasphere content. This plasmasphere temporal pattern is similar to that displayed in the SUPIM simulations and Taiwan data presented by Mazzella et al. [2007]. Of further note is the smoothness of the ionospheric temporal profile throughout the high and middle latitudes, and even for the temporal segments of the southernmost stations not influenced by the equatorial anomaly crest. This feature was previously noted by Mazzella [2009] for the IGS station at Westford, where significant latitudinal TEC gradients appear if the PEC is included in the composite equivalent vertical TEC. For this chain of stations, examples of this latitudinal gradient effect are displayed in Figure 7 for Neah Bay and La Paz, which correspond to the SCORE depictions of the ionosphere (but using SCORPION biases).

Figure 6a.

The SCORPION results, without plasmasphere baseline adjustments, for the poleward half of the 16 stations in Table 1.

Figure 6b.

The SCORPION results, without plasmasphere baseline adjustments, for the equatorward half of the 16 stations in Table 1.

Figure 7.

SCORE depictions of the composite (ionosphere plus plasmasphere) equivalent vertical TEC results for the Neah Bay, WA, and La Paz, Mexico, stations, for the same SCORPION calibrations as in Figures 6a and 6b, displaying a larger variability in the equivalent vertical TEC (EQ VER TEC) associated with the latitude of the IPP. Note that the SCORE convention of using the geographic IPP local time (IPP GEO LOCAL TIME) and latitude (GLAT) is employed here, in contrast to the magnetic IPP coordinates used in Figures 6a and 6b.

[15] Figures 8a and 8b present the pairwise comparisons for the plasmasphere baseline determination in a graphical form, prior to the application of any plasmasphere baseline corrections. The equivalent vertical IEC is also displayed, for assessment of the calibration accuracy (including possible effects from a variable effective altitude for the slant-to-vertical conversion, especially near the latitudinal extremes of the chain of stations). The poleward view from Holberg (HOLB), Canada, is notable for having a distinctly low IEC during the night, but this appears to be a bias determination error, possibly arising from an inappropriate slant TEC conversion factor or incorrect partition of TEC between the ionosphere and plasmasphere. However, the latter possibility appears unlikely because of the agreement in PEC between Level Island (LEV2), Alaska, and Holberg, and Holberg and Neah Bay, as well as the IEC agreement between Holberg and Neah Bay (Figure 8a and Table 2).

Figure 8a.

Pairwise comparisons for the plasmasphere baseline evaluation, prior to any plasmasphere baseline adjustments, poleward of Petaluma, CA (P198). The top plot in each panel displays the vertical PEC, as in Figure 4, and the TEC scale for the ionosphere changes progressively along the chain. For each comparison, the poleward station is displayed in red and the equatorward station is displayed in blue.

Figure 8b.

Pairwise comparisons for the plasmasphere baseline evaluation, prior to any plasmasphere baseline adjustments, equatorward of Petaluma, CA (P198). The top plot in each panel displays the vertical PEC, as in Figure 4, and the TEC scale for the ionosphere changes progressively along the chain. For each comparison, the poleward station is displayed in red and the equatorward station is displayed in blue.

Table 2. Pairwise Comparisons for the Plasmasphere Baseline Determinationa
Site IdentifierAverage PSREP VTEC (North)Average PSREP VTEC (South)Delta PSREP VTECPlasmasphere Baseline
  • a

    Note that the successive Plasmasphere Baseline differences are distinct from the pairwise plasmasphere vertical TEC differences (Delta PSREP VTEC) because of the intrinsic distinction between the two quantities, which correspond to slant and vertical TEC measurements, respectively.

FAIR 0.056 0
COL22.301 −0.1530.53

[16] Table 2 presents the pairwise comparisons for the plasmasphere baseline determination in a numerical form, based on the data associated with Figures 8a and 8b. Most of the individual inter-station differences are small (less than 0.1 TEC unit in absolute value), and nearly half of the differences are negative, arising from the limitations in the accuracy of the individual plasmasphere determinations. Only the baseline corrections for Point Loma, CA, to Colima, Mexico, which are positive and exceed 0.2 TEC units, were applied. The Ap index reached 30 and the Dst index decreased to −63 on 1 April 2007 (day 91), indicating some possible disruption of the plasmasphere, so the plasmasphere on 8 April 2007 could still be in the replenishment phase, and the lack of a significant plasmasphere baseline appears reasonable.

[17] The nominal receiver biases (defined as the average of all of the composite satellite-plus-receiver biases) for each station are presented in Figure 9, with distinct symbols identifying the receivers for which the GPS P1 (precise code) signal was available from those for which the C1 (coarse acquisition code) signal was used because of the absence of P1. Also shown are the receiver biases derived by Center for Orbit Determination in Europe (CODE) and Jet Propulsion Laboratory (JPL) for Fairbanks and Holberg. The SCORPION error bars are derived from the normalized, square-root of the quantity (E) presented in equation (1), while the CODE and JPL error values were provided with the respective bias values. The derived satellite biases, relative to the mean bias for each receiver, are displayed in Figure 10 separately for the receivers reporting P1 and those not reporting P1. Although fewer of these receivers report P1, the standard deviation for their individual relative satellite biases is still considerably smaller (<1.1 TEC units) than for the receivers requiring C1 usage (up to 2.6 TEC units, and typically greater than 1.2 TEC units). The relative satellite biases for Glennallen (GNAA), Alaska, are somewhat discrepant from the other receivers in its group, and more closely resemble the relative biases for the P1 usage receivers.

Figure 9.

The nominal receiver biases for each station, in TEC units, with squares for the stations allowing GPS P1 processing and circles for the stations where the GPS C1 processing was required. For comparison, the CODE (upright triangles) and JPL (inverted triangles) results, from the respective IONEX files, are also displayed (FAIR and HOLB).

Figure 10.

The derived satellite biases, relative to the mean bias for each receiver, for each station, in TEC units, separately for the (a) C1 and (b) P1 processing.

5. Comparisons to Other Data Sources and Methods


[18] Because SCORPION was developed to overcome the limitations of the SCORE algorithm with regard to the plasmasphere [Mazzella et al., 2002], an obvious comparison between SCORPION and SCORE presents itself for this chain of stations. Using the same data provided as input to SCORPION (with the multipath corrections applied), calibrations were performed by the SCORE method. This was implemented here as a full SCORE emulation by SCORPION, including the original use of geographic coordinates to define the IPP locations, and the standard SCORE parameters to define the conjunctions, weight factors, and data selection region, while also suppressing the plasmasphere determination. The differences between the mean biases over all satellites (i.e., the nominal receiver biases) at each station are displayed in Figure 11. From Fairbanks (FAIR) to Neah Bay (NEAH), the derived SCORE bias decreases relative to SCORPION, becoming increasingly negative, while equatorward of Neah Bay, the derived SCORE bias increases relative to SCORPION, eventually producing a positive bias difference in the vicinity of La Paz (LPAZ) and Aguascalientes (INEG).

Figure 11.

The nominal receiver bias differences between SCORE and SCORPION for each station, in TEC units, showing the plasmasphere influence on the SCORE calibrations. The error bars are composites of the SCORPION and SCORE error values.

[19] As demonstrated by Mazzella et al. [2007] and also noted by Anghel et al. [2009], the effect of uncompensated PEC on TEC calibrations is to produce an underestimate of bias values (and higher TEC values) for middle latitude regions while producing an overestimate of bias values (and lower TEC values) for equatorial regions. Their reported effects for the middle latitude regions confirm the initial investigations and validations by Lunt et al. [1999b] (near 50° magnetic latitude), while the effects for equatorial regions were previously noted at least as early as a 1998 observation campaign at Ascension Island [Fremouw et al., 1998], where the overestimate of the biases was significant enough to produce negative TEC estimates. The latter phenomenon also appears in the Kwajalein data of Lin et al. [2007] using a calibration method different from SCORE.

[20] The bias comparison between SCORPION and SCORE for this chain of stations is also supported by calibrations using simulated data (with known zero biases and no multipath errors) derived from SUPIM, for a small set of stations along the chain (Figure 12). Although the details of the simulated data (and derived biases) differ from the measured GPS data, the same latitudinal trend in the biases derived by SCORE is evident, while the SCORPION biases display little significant trend.

Figure 12.

Separate SCORE and SCORPION receiver biases derived for SUPIM simulations. The SCORPION biases have been adjusted to compensate for the plasmasphere baseline corrections.

5.2. SCORPION Versus LEO Tomography

[21] The locations of the three GPS stations in Alaska allow comparisons to the TEC results determined by the chain of tomography receivers associated with the High-frequency Active Auroral Research Program (HAARP) [Andreasen et al., 2004]. These receivers use dual-frequency signals from Transit satellites and other LEO satellite transmitters to derive TEC measurements up to the Transit or LEO satellite altitudes, essentially measuring only the ionospheric contribution to TEC. Figure 13 displays a comparison of SCORPION IEC measurements to the tomography TEC results, for several Transit satellite passes. The SCORPION IEC values are 1–2 TEC units larger than the tomography TEC results, which is somewhat larger than the combined error of about 0.6 TEC units.

Figure 13.

A comparison of the SCORPION IEC (GPS) measurements to HAARP tomography TEC results, for several LEO satellite passes.

5.3. SCORPION Versus Jason-1

[22] As noted by Bishop et al. [2009], a Jason-1 pass segment (cycle 193, pass 130) is suitably situated to provide comparative data for the GPS TEC measurements on this day. This pass segment commences at 01:26:30 UT in the vicinity of Cape Hinchinbrook (CHI4), Alaska, (see Figure 1) providing a Jason-1 calibration opportunity against a GPS station that is not only coastal but also in a region of minimal plasmasphere content. This calibration was performed by SCORPION, treating the GPS TEC data as a known reference from the previous SCORPION calibration and the Jason-1 TEC data as uncalibrated, using the multiple-station capabilities of SCORPION. To insure the exclusion of PEC from the GPS measurements, the Jason-1 TEC measurements were calibrated against only the IEC determined by SCORPION for Cape Hinchinbrook. The derived Jason-1 bias was 4.62 TEC units, with an additional limit of up to 0.33 TEC units possible from the vertical PEC in this region. However, there is an initial transient in the Jason-1 TEC (see Figure 14), and this presents a larger contribution to the uncertainty in the Jason-1 bias than the plasmasphere. An alternative GPS reference station at Kodiak Island (KOD2), Alaska, was investigated for the Jason-1 calibration, but was also confronted by the same Jason-1 TEC transient. However, the Jason-1 (sub-satellite) IPPs did satisfy conjunction criteria with GPS PRN 25 as observed from Level Island (LEV2), Alaska, for a somewhat less variable segment of the Jason-1 pass, so this site was used for the Jason-1 TEC calibration, producing a Jason-1 bias value of 5.60 TEC units. The error bars associated with the Jason-1 values were derived solely from the standard deviations for the averaged data shown in Figure 14, over an averaging interval of about 66 s, and do not include any estimates for the Jason-1 bias error.

Figure 14.

Latitudinal comparison of Jason-1, SCORE, and SCORPION TEC values. (bottom) The equivalent vertical TEC for the ionosphere only; (top) the vertical PEC at the IPP from SCORPION; and (middle) the composite vertical TEC from SCORPION, with the same Jason-1 and SCORE TEC as in Figure 14 (bottom).

[23] For Figure 14, the data segments associated with each GPS station were selected as one-hour intervals in magnetic local time bracketing the (local) Jason-1 pass segment, with an additional restriction of lying within 7.5° of the site geographic longitude. (The gap in comparisons around 35° latitude arose from the delayed occurrence of the first complete GPS pass for the associated stations YBHB, P198, and P067 for the specified 8 April 2007 data period, relative to the passage of Jason-1. Likewise, no comparisons occurred for HER2 and INEG.) Figure 14 (bottom) displays the equivalent vertical TEC for the ionosphere only, although for somewhat incommensurable measurements from Jason-1, SCORPION, and SCORE, because the SCORPION ionospheric ceiling altitude is 1000 km, the Jason-1 ionospheric ceiling altitude is 1336 km (the satellite altitude), and the SCORE ionospheric ceiling altitude is 20190 km (the GPS satellite altitude). However, SUPIM calculations indicate that the SCORPION ionosphere and Jason-1 values should agree to within about one TEC unit. Figure 14 (top) displays the vertical PEC at the IPP, as described for Figure 3, so this quantity is available only from the SCORPION results. Figure 14 (middle) displays the composite vertical TEC, which is the sum of Figures 14 (bottom) and 14 (top) for the SCORPION results and identical to Figure 14 (bottom) for the Jason-1 and SCORE results.

[24] For the vertical IEC comparisons, the SCORE values are equal to or greater than the Jason-1 values, which would be expected from their relative coverages in altitude, although the expected trend in the difference does not appear to be maintained at the lowest latitudes, which is a possible corroboration of the bias error trend noted for Figure 11. The SCORPION IEC values are comparable to the Jason-1 values, although sometimes slightly lower than the Jason-1 values over the middle latitude range (40°–56°). An unresolved issue is whether this is a consequence of the initial transient for the Jason-1 pass, affecting its bias estimate, a systematic difference in the SCORPION calibrations between middle (40°–56°) and high (above 56°) latitudes, or a residual variance in the Jason-1 values. The comparison between SCORPION and Jason-1 appears somewhat better for the composite vertical TEC, although the Jason-1 values still sometimes exceed the SCORPION values at middle latitudes, which should not occur.

[25] Two alternative calibrations of the Jason-1 TEC are also possible for this pass segment. One method is the direct analog of the SCORPION GPS cross-calibration, but using the SCORE calibration results (and presuming a negligible plasmasphere effect). The second method is a comparison of the Jason-1 data to the HAARP tomography TEC values. However, this comparison is affected by the same Jason-1 TEC transient that precluded the cross-calibrations using the GPS stations at Cape Hinchinbrook and Kodiak Island, so this method was not attempted for this study.

[26] For the Jason-1 cross-calibration against the Level Island SCORE GPS TEC, the additional provision of a lower latitude exclusion was incorporated for the Level Island SCORE GPS TEC calibration, based on the SCORE/SCORPION bias differences reported in Figure 11 and the previous study by Lunt et al. [1999b]. Only data for IPP geographic latitudes above 55.966° (0.5° south of the station) were used for the SCORE GPS TEC calibration, producing an average bias that was only 0.03 TEC units greater than the derived SCORPION average bias for that station, in contrast to the difference of −1.31 TEC units displayed in Figure 11, derived without a latitudinal exclusion. However, the resultant Jason-1 calibration against the GPS TEC values derived from SCORE was 3.14 TEC units, influenced primarily by the residual PEC in the SCORE GPS TEC values. As noted by Bishop et al. [2009], a poleward viewing perspective for the GPS observations would ameliorate this situation, but the nature of this Jason-1 pass segment and the GPS coverage by this chain preclude even moderately poleward viewing perspectives for any of these stations north of Fort Stevens. An alternative Jason-1 pass segment and different GPS station selection could resolve this issue for SCORE calibrations (or similar methods that do not account for the PEC), but the circumstances are not so restrictive for SCORPION, because of its partitioning of TEC between the ionosphere and the plasmasphere.


[27] Various analysis centers provide GPS-derived TEC maps in the Ionosphere Map Exchange (IONEX) format, permitting another comparison for these SCORPION results, although without separate contributions for the ionosphere and the plasmasphere. For this study, the maps generated by the Center for Orbit Determination in Europe (CODE) of the Astronomical Institute of the University of Bern, the Jet Propulsion Laboratory (JPL) of the California Institute of Technology, and the Research group of Astronomy and Geomatics of the Technical University of Catalonia (UPC) were used. The UPC map is notable for utilizing two layers for the GPS TEC analysis, centered on altitudes 399 km and 1079 km (which could be associated with the ionosphere and plasmasphere, respectively), and therefore has some generic similarities to SCORPION, but also many differences. Furthermore, the portrayal of their results is somewhat restricted by the single-layer representation used for the IONEX files, although this restriction is not intrinsic to the IONEX format [Schaer et al., 1998].

[28] Figures 1518 display the latitudinal profiles of vertical TEC for each of the three IONEX maps and both SCORE and SCORPION, for four magnetic local times (−06:00, 00:00, 06:00, and 12:00). The IONEX map data were selected along a local geographic meridian for each site at each magnetic local time, while the GPS data were selected using the corresponding magnetic local time and longitude criteria as for the Jason-1 comparisons. The apparent contravening latitudinal gradients for some of the GPS data segments (where the TEC increases with increasing latitude instead of decreasing) are a consequence of the temporal variation of TEC overwhelming the normal latitudinal gradient effect evident for the IONEX maps.

Figure 15.

Latitudinal comparison of IONEX, SCORE, and SCORPION TEC values, in the same format as Figure 14, for a nominal magnetic local time of −06:00 (18:00 for the previous day). IONEX results from the processing centers CODE, JPL, and UPC are shown, with no distinction for these results between the ionosphere values (Figure 15, bottom) and composite ionosphere and plasmasphere values (Figure 15, middle). Note the appearance of contravening TEC latitudinal gradients (prominently for 20°–30°) for the SCORE and SCORPION data, arising from the temporal variation of TEC at this local time. (See Figures 6a and 6b.)

Figure 16.

Latitudinal comparison of IONEX, SCORE, and SCORPION TEC values, of the same type as Figure 15, but for a nominal magnetic local time of 00:00. Note the reduced apparently anomalous latitudinal TEC gradients at the lowest latitudes (20°–30°), compared to Figure 15, associated with the smaller temporal TEC variations at this local time. (See Figures 6a and 6b.) Also note the general continuity of the SCORPION ionosphere vertical TEC values (Figure 16, bottom) compared to the “sawtooth” pattern for the corresponding SCORE values, especially for latitudes between 30° and 50°.

Figure 17.

Latitudinal comparison of IONEX, SCORE, and SCORPION TEC values, of the same type as Figure 15, but for a nominal magnetic local time of 06:00. Note the reappearance of the contravening latitudinal gradients, again arising from the temporal variation of TEC at this local time.

Figure 18.

Latitudinal comparison of IONEX, SCORE, and SCORPION TEC values, of the same type as Figure 15, but for a nominal magnetic local time of 12:00. The “sawtooth” pattern for the SCORE TEC values is again apparent for the stations where the temporal variation of TEC is unimportant (40°–65° latitude). (See Figures 6a and 6b.)

[29] The comparisons between SCORPION and SCORE for these four magnetic local times generally display the same variation with latitude as in Figure 14 except at the lower latitudes, where the SCORE equivalent vertical TEC remains larger than the SCORPION (ionosphere) equivalent vertical TEC, but becomes only approximately equal to the SCORPION composite vertical TEC, indicating the effects of the plasmasphere for the respective methods, especially when considered in association with the bias differences (Figure 11).

[30] The CODE IONEX TEC values tend to be larger than those for UPC at the lower latitudes of this chain, while the reverse occurs for the higher latitudes, and the JPL IONEX TEC values are larger than either of these throughout this latitude range, as well as generally being larger than the SCORE and SCORPION values. Both the CODE and UPC IONEX TEC values tend to be closer to the SCORE equivalent vertical TEC than to the SCORPION composite vertical TEC, but this may be more a consequence of the standard determination of equivalent vertical TEC from slant TEC than an influence of the bias determinations.

[31] Figure 19 displays a variant of Figure 18, with the SCORPION results (labeled as “Emulation”) in Figures 19 (middle) and 19 (top) presented as an equivalent vertical electron content converted from the slant electron content using the slant conversion factor applied for the ionosphere. Thus, the SCORPION results in Figure 19 (middle) correspond directly to the SCORE results, differing only by the derived biases. In contrast to Figure 18, which displays a relatively smooth vertical total electron content (VTEC) latitudinal gradient for the middle latitude SCORPION results, Figure 19 displays a “sawtooth” VTEC pattern for the SCORPION results very similar to the SCORE results. Likewise, the plasmasphere equivalent vertical TEC in Figure 19 displays a similar “sawtooth” pattern instead of the smooth, continuous gradient of Figure 18, clearly identifying the plasmasphere as the source of this pattern. This pattern also corresponds to the plasmasphere effect noted by Mazzella [2009] from SUPIM simulations for derived equivalent vertical TEC.

Figure 19.

SCORPION emulation of SCORE conversion from slant TEC to equivalent vertical, for (bottom) the ionosphere only; (middle) the composite ionosphere and plasmasphere; and (top) the plasmasphere only, indicating the plasmasphere role in producing the “sawtooth” pattern displayed in the SCORE results.

6. Further Discussion and Conclusions

[32] The SCORPION technique provides a method for GPS bias calibrations that also determines values for PEC, both along the line-of-sight and vertically at the IPP, which are generally distinctly separated regions for even moderate elevation angles away from local vertical [Mazzella, 2009]. The SCORPION determination of vertical PEC at the IPP also resolves the apparent “sawtooth” irregularity for the composite IEC and PEC derived by SCORE, arising from the mapping and conversion of distant slant PEC to the IPP, thus compressing the PEC gradient from a gradual, smooth variation into a more steeply varying quantity.

[33] The resolution of the plasmasphere baseline described here removes the potential step discontinuities in the vertical PEC latitudinal gradients, and also resolves the ambiguity between the plasmasphere baseline and the receiver bias, improving the determination of the composite vertical IEC and PEC. This plasmasphere baseline determination is also a consideration for other methods that partition slant TEC into ionosphere and plasmasphere contributions when the plasmasphere contributions are determined from the GPS measurements [e.g., Anghel et al., 2009; Carrano et al., 2009].

[34] The bias determinations by SCORE and SCORPION for simulated data confirm the latitudinal trend in bias errors previously reported by Mazzella et al. [2007] and Anghel et al. [2009], while the correspondence between this trend and the bias differences between SCORE and SCORPION supports the benefit of the latter method for GPS calibrations. These results also contradict the assertion by Bishop et al. [2009] that a monotonic trend in bias errors for their subset of stations is improbable, rather than an expected consequence of PEC upon SCORE.

[35] The bias error trend associated with the plasmasphere latitudinal variation has implications for the determination of PEC using differential ground-based GPS TEC measurements [Lunt et al., 1999d; Law, 1999]. For a pair of latitudinally separated stations, as in Figure 3, the equatorward station would have a positive bias error relative to the poleward station, so the poleward lines-of-sight from the equatorward station through common IPPs would have a systematically reduced slant TEC value, leading to an increased differential slant TEC measurement. If this differential slant TEC is attributed to the PEC, then the PEC will be overestimated. The studies by Lunt et al. [1999d] and Law [1999] employed latitudinal restrictions for the SCORE calibrations to reduce this systematic bias error, but the study by Bishop et al. [2009] appears to have omitted such provisions, which would affect their quantitative results.

[36] Separately from comparisons to SCORPION, the influence of the plasmasphere on calibration methods other than SCORE may be discernable by the correlation of day-to-day bias variations against PEC, with the latter being determined by either the GPS minus LEO TEC comparisons [Ciraolo and Spalla, 1997; Lunt et al., 1999c] or the two-station GPS TEC comparisons [Lunt et al., 1999d; Law, 1999], or indirectly represented by the number of plasmasphere replenishment days since a storm [Lunt et al., 1999c]. A cyclic bias variation on an annual basis may also be indicative of plasmasphere influences, as individual satellite passes migrate through the nighttime period, when the PEC is larger relative to the IEC, especially for low solar flux conditions [Lunt et al., 1999a; Mazzella, 2009].

[37] The low standard deviation associated with the relative satellite biases using P1 measurements (section 4) somewhat supports the effectiveness of the multipath treatment used for this study. One consequence of this is the possible use of relative satellite biases for the SCORPION bias and plasmasphere parameter determinations, using provisions already incorporated into SCORPION, with the expectation of improving the reliability of the receiver bias and plasmasphere parameter determinations by reducing the number of degrees of freedom.


[38] The author is grateful to Graham J. Bailey for providing SUPIM. Development of SCORPION was conducted in collaboration with G. Susan Rao and supported by the Air Force Research Laboratory Space Vehicles Directorate under SBIR contracts FA8718-04-C-0009 and FA8718-05-C-0026 to NorthWest Research Associates. The GPS data were obtained from the United States Department of Commerce, National Oceanic and Atmospheric Administration (NOAA) FTP site for the CORS stations [Noll, 2010] and from the Crustal Dynamics Data Information System (CDDIS) at the NASA Goddard Space Flight Center FTP site for the IGS station (HOLB) [Dow et al., 2009]. The HAARP tomography data were obtained from the HAARP Web site The Jason-1 data were obtained from the Physical Oceanography Distributed Active Archive Center of the Jet Propulsion Laboratory FTP site [Berwin, 2003]. The IONEX data were obtained from the Crustal Dynamics Data Information System (CDDIS) at the NASA Goddard Space Flight Center FTP site [Dow et al., 2009], and the associated processing code was obtained from the University of Berne FTP site [Schaer et al., 1998]. The Ap indices were obtained from the NOAA National Weather Service Space Weather Prediction Center Web site The Dst indices were obtained from the World Data Center for Geomagnetism, Kyoto Internet site The figures were prepared using the Generic Mapping Tools (GMT) graphics [Wessel and Smith, 1998].