Details of the data acquisition and data processing techniques were presented in previous papers [Morabito et al., 2003; Morabito, 2007]. Detail that is relevant to the current study is repeated here. The data used for the analysis of spectral broadening were acquired on Full-Spectrum Receiver (FSR) open loop receivers. The complex signal was rotated using a polynomial fit to remove long-term trends due to unmodeled Doppler and residual spacecraft motion. The phase corrected signal was analyzed with a software phased locked loop yielding measurements of SNR, frequency and phase from which the parameters and their statistics were estimated. The power spectrum of the complex signal was analyzed with a Fast Fourier Transform (FFT) script file. The broadened bandwidth, B, was estimated from the power spectrum averaged over successive 400 s periods. The PLL output SNR was used to estimate the amplitude scintillation index, m. The PLL output frequency time series (with long-period nonsolar trends removed as described above) was used to estimate phase fluctuations.
4.1. Phase Scintillation
 Estimates of received signal frequency extracted from the PLL processing can be used to deduce plasma phase scintillation when solar effects dominate at the appropriate time scales. One method produces estimates of received frequency residuals using software PLL processing on each individual band of FSR data where long period (nonsolar) frequency trends were removed. The RMS scatter of the resulting frequency data was found to increase markedly at both bands as the SEP angle decreased [Morabito et al., 2003], where the X band scatter always exceeded the level of the Ka band scatter by the expected ratio. Examples of Allan deviation curves for selected time scales at both X band and Ka band were previously reported on [Morabito et al., 2003] for cases of SEP = 3.1° and SEP = 0.6°. The phase fluctuations over all time intervals at SEP = 0.6° were significantly elevated relative to those observed at SEP = 3.1° as expected. Figures 2a and 2b display the Allan deviation signatures for the ingress and egress Cassini 2000 solar conjunction passes (excluding the 0.6° passes) respectively. The Allan deviation trends with time scale are roughly Kolmogorov (consistent with ∼ τ−1/6 dependence in equation (5)), with flatter examples such as 2000/130 (Figure 2a) corresponding to steeper spectra. These Allan deviation signatures all show a decrease at small 1 to 3 s time scales that correspond to the high-frequency cutoff of the spatial spectrum of the density irregularities. These time scales are consistent with typical solar wind velocities encountered and ∼100 km upper bound of the inner scale of turbulence. Energy below these time scales is dissipated where the corresponding spatial spectrum becomes very small.
Figure 2. Allan deviation versus time interval of X/Ka differenced frequency for each pass (solid curves) of (a) ingress and (b) egress. Pass ID and SEP angle are annotated on each curve along with corresponding impact distance in RS. Also shown for reference is the model Allan deviation dependence for Kolmogorov turbulence (dashed curves). (c) Allan deviation values at 1000 s from Figures 2a and 2b as a function of solar radii along with model (see text).
Download figure to PowerPoint
 It can be seen that the Allan deviation curves progress to noisier levels as the SEP angle between Cassini and Sun decreases during ingress (Figure 2a) and to less noisier levels as the SEP angle increases during egress (Figure 2b). The Allan deviation measurements at τ = 1000 s were extracted from Figures 2a and 2b and plotted in Figure 2c as a function of solar radii. The measurements ranging from 10−10.5 to 10−12 at the smallest SEP angles are generally consistent with those of other experimenters [Asmar et al., 2005; Armstrong, 2006]. The model of Armstrong et al.  (scaled appropriately to X band) is also provided for comparison purposes in Figure 2c (solid curve).
 System noise contributions on these data were found to be negligible [Morabito et al., 2003]. Allan deviation is typically used by spacecraft mission designers to ascertain system performance degradation on phase at different SEP angles for a variety of telecom and navigation scenarios, including radio science experiment noise budgets. The phase structure function is used by turbulence theorists to measure density variations of various media. Given that the Allan deviation and phase structure function are essentially linearly related to each other, the reader is referred to Figure 13 for selected examples of phase structure functions extracted from these data.
 The Allan deviation curves at the smallest SEP angles are expected to exhibit a larger variance depending upon the solar activity along the signal path. At the smallest observed SEP angle of 0.6° during this solar conjunction (Allan deviation signatures not shown in Figures 2a and 2b), significant nonstationary changes in the frequency residuals due to solar coronal activity can significantly skew the Allan deviation estimates. Evidence of significant structure in the frequency residual time series at SEP = 0.6° during 2000/133 is evident in Figure 3 during a very turbulent 24 min period that spanned from 2000/133 23:48 to 2000/134 00:12 UTC. The large variations in frequency are presumably due to a sizable solar event that was seen in white light images hurling huge clouds of material outward into space that traversed the signal path between Cassini and the receiving station. Positive Doppler shifts are caused by material approaching the line of sight signal path in the transverse direction while negative Doppler shifts are caused by material receding away from the line of sight after passing through it. The variations in frequency at X band (Figure 3a in red) and Ka band (Figure 3a in blue, offset slightly for illustration purposes) are well correlated and their relative levels are consistent with those expected due to charged particle inducements. The differenced frequency residual signature [(fX − fKa/3.8)/(1–3.8−2)] displayed in Figure 3b characterizes the variations due to plasma only.
Figure 3. Frequency residuals (1 s) from PLL processing of open loop receiver data for 2000/133 23:48 to 2000/134 00:12 UTC at SEP = 0.6° for (a) X band (red curve) and Ka band (blue curve) and (b) X/Ka difference.
Download figure to PowerPoint
 Using the model of equation (1), the electron number density fluctuation estimated at the various impact distances of the raypath were extracted from the Doppler noise measurements, based on the geometric parameters of this solar conjunction. It was found that the resulting estimates of fluctuation in electron number density lay within about 1% of the total electron number density expected at the different solar impact distances using a modified Allan-Baumbach formula [Bird et al., 1994] for the quiet background. For example, at 4 RS impact distance, the estimated fluctuation in electron number density runs ∼300/cm3. The total electron number density at 4RS ranges from ∼104/cm3 (3% fluctuations) to 1.4 × 105/cm3 (0.2% fluctuations) depending upon values used for the model coefficients and other assumptions (solar latitude and solar cycle phase). Assuming a solar latitude of near −90° during solar maximum conditions using a solar latitude dependent model by Muhleman and Anderson , this fluctuation translates to about 1% of the total electron number density, assuming nontransient background conditions. This level of electron density fluctuation was also found to be consistent with an estimate derived from an adjustment of an integrated model along the signal path used to match amplitude scintillation measurements [Morabito, 2007]. During periods of enhancements, the increased phase noise will translate to density fluctuations of several times this value.
 Figure 4a displays a longer period of X band frequency residual data than what was shown in Figures 3a and 3b. Figure 4b displays column density that was derived from the data of Figure 4a. As these data only provide information on relative column density changes, the absolute level was set by forcing the minimum value during this period to be equal to the expected quiescent background column density (2.9 × 1020/m2), estimated using the electron density model of Muhleman and Anderson  with an integration along the radio raypath using Cassini solar conjunction geometry at a minimum solar impact distance of 4.2 RS. Given that the absolute level of column density in Figure 4b may be unknown, what is important is that the change in column density can be estimated by differencing the values between any two times. This difference can then be compared with any given quiescent background level. The difference between peak and minimum value in Figure 4b is ∼2.0 × 1021/m2 which is about three times larger than the assumed background column density expected at this solar impact distance. Woo and Armstrong  noted an increased factor of 3.7 in peak electron density over that of a preshock value using similar techniques, which was found to be within expectations. The actual background level at the Cassini raypath impact point may likely be higher given the degree of turbulent activity over much of this period. The background column density could be estimated by using inversion techniques on the SOHO image brightness for purposes of correlation with the radio data, and this is a subject of further study. Radio measurements such as those described in this paper could also provide a useful calibration in the extraction of column density from white light images.
Figure 4. (a) Frequency residuals (1 s) from PLL processing of X band open loop receiver data for 2000/133 23:30 to 2000/134 00:54 UTC at SEP = 0.6°, (b) column density derived from X band frequency residuals with arbitrary setting of minimum value (see text), and (c) concurrent spectral broadening measurements acquired during same period (X band upper curve, Ka band lower curve), where smoothed curves connect data points.
Download figure to PowerPoint
 Since the available X band data shown in Figure 4a covers a longer period than the time span of the valid Ka band data shown in Figure 3a, the longer arc of X band only data shown in Figure 4a was used for estimating the column density signature in Figure 4b as previously discussed. The similarity of the X band only frequency residual signature shown in Figure 3a and that of the plasma only component shown in Figure 3b suggests that the X band only data should suffice for the plasma analysis of the longer data arc. The dual band frequency residuals (fX − fKa/3.8)/(1–3.82) shown in Figure 3b were converted to phase and then to column density as a function of time using the appropriate formulation. The column density time series was similarly scaled as described above. The signature of the column density changes extracted from the differenced data from Figures 3b was then compared to the signature of the column density of the common period determined from the X band only data shown in Figure 4b. These two curves agreed well suggesting that the estimation of the column density from the wider span of X band only data is valid. This is also intuitively justified given that the fluctuations in the X band only time series are clearly dominated by solar charged particles, as the other error sources are negligible, except possibly for error induced by the long-term trend removal steering of the raw data. An error in the estimated coefficients derived from the frequency steering algorithm used to remove the long period trends due to trajectory, troposphere, etc. could change the derived column density signature and thus skew the difference between the minimum and maximum values used to estimate column density change during the event. An assessment of frequency steering error was made by generating additional column density curves using minimum and maximum slopes based on the error bars in the frequency steering. The resulting error on the column density change projected from using these curves was found to be at the 10% level. The resulting Figure 4b column density signature is thus validated using the X band only Doppler residuals shown in the top panel of Figure 4a.
 Figure 4c displays the spectral broadening measurements acquired during this same period at both X band and Ka band. Note that the dominant features near 2000/133 23:54 UTC almost coincide with the dominant phase features shown in Figure 4a. Figure 4b displays the column density during the passage of a CME during 2000/133 that took about 35 min to traverse the signal path. Assuming a velocity of 400 km/s, this translates to an extent of the CME “cloud” measuring about 840,000 km. A rough calculation of the number of electrons assuming that the CME is roughly spherical in shape is about 1039 particles. This corresponds to an electron mass of ∼109 kg or about ∼1012 kg if we account for the mass of associated protons. If we assume the linear speed of 2604 km/sec listed in the CME Catalogue for this event, the equivalent proton/electron mass estimate becomes 5 × 1013 kg (50 billion metric tons), which is a mass comparable to that of the most massive CMEs that have been observed.
4.2. Spectral Broadening
 Figure 5 provides a visual interpretation of spectral broadening bandwidth (enclosing half of the area of the RF carrier power spectrum) versus SEP angle for the Cassini 2000 solar conjunction. Significantly more detail is provided in this plot over that originally presented in Figure 12 of Morabito et al. , where only a small subset of selected quiet period data were shown. Here in Figure 5, the full series of B measurements are plotted for each of the smallest SEP angle passes (∣SEP angle∣ < 2°) of the Cassini 2000 solar conjunction. For pass 2000/137 near +2.4 deg SEP, only the average value of broadened bandwidth is plotted in Figure 5 for both X band and Ka band. Along with the data points in Figure 5, the models for X band (solid red curve) and Ka band (dashed blue curve) are also displayed, which nicely traverse the quiet period data points for each pass (where minimum bandwidth points reside). The bandwidth measurements lying above the model represent periods where solar transient activity was known to have occurred. By temporally analyzing these measurements separately for each pass during the maximum of Solar Cycle 23 (in the discussion to follow), we see that transient activity can be easily detected above background levels.
Figure 5. Spectral broadening bandwidth versus SEP angle for selected Cassini 2000 solar conjunction passes. Individual data points represent 400 s periods where the carrier spectrum was processed to extract the bandwidth over which half of the power resides. Red triangles are X band data points. Blue diamonds are Ka band data points. Solid red curve is X band model. Dashed blue curve is Ka band model.
Download figure to PowerPoint
 The X band and Ka band model curves plotted in Figure 5 were obtained using a “fixed p” model which depends on solar conjunction geometry, a density fluctuation profile for Cn2, a solar wind velocity profile for v, and power law index, p [Woo et al., 1977] as given by equation (8). The quiet background levels of measurements of B acquired during this period of solar maximum in the region below the south solar pole are in good agreement with the model. The numerical values used for the Cn2 profile, the velocity profile v and power law index p (3.5) were the same as used in the amplitude scintillation model [Morabito, 2007]. Thus the quiet model background curves for X band and Ka band carrier bandwidth broadening are consistent with the electron number density fluctuation level (300/cm3 at 4 RS) extracted from the amplitude scintillation study [Morabito, 2007]. A slightly different value of p can be used in the model to better match the minimum values of the observations.
 The two smallest SEP angle (0.6°) passes (2000/133 and 2000/134) occur as the spacecraft signal source in the sky (impact points) pass through the southern polar region of the solar environment as shown in Figure 1. Given that the model was based predominately on ecliptic region measurements, these results are consistent with the suggestion of Woo and Goldstein  that latitudinal variation of solar mass flux may be weaker during solar maximum.
 The magnitude of the X band broadened bandwidth is typically a factor of five times that of the concurrent Ka band broadened bandwidth measurement (equation (10)), when it is assumed that the density spectrum is isotropic and there are no velocity fluctuations. It is instructive to compare the measured ratio with the theoretical relationship. Thus, the time series of the measured ratio (BKa/Bx) was estimated from the data and the statistics extracted for each pass. In all cases the measured bandwidth ratio agrees quite well within the range of expected BKa/Bx ratios from 0.11 to 0.2. There are also trends present in the data where this ratio varies between ∼0.15 and ∼0.2. Scatter and changes in this ratio are attributed to a combination of statistical noise as well as divergence from model assumptions, such as the presence or lack of velocity fluctuations, and turbulent changes in electron number density. The changes in this bandwidth ratio can in turn be related to changes of the power law exponent (see equation (11)).
 Table 1 presents a statistical summary of the spectral broadening observations for both X band and Ka band, the statistics of the individual time series ratios of the dual-band broadening measurements, and the statistics for the estimated power law index using equation (11). For all but one pass, these are presented separately for both quiet periods and active (transient) periods. For several cases, the results are statistically consistent with the Kolmogorov power law index of p = 3.67. The mean and 1σ scatter for the active period bandwidth measurements shown in Table 1 are much higher than those of the quiet periods as they are indicative of increased turbulence in the form of increased velocity and density variations during these periods. The quiet period values of broadened bandwidth at both frequency bands are in good agreement with the model values (shown as the curves in Figure 5) at the respective SEP angles.
Table 1. Statistical Summary of Spectral Broadening Results
|Year/Day||SEP (deg)||Data Selection||BX (Hz)||BKa (Hz)||BX/BKa||p|
|2000/132||1.1||Quiet||1.30 ± 0.35||0.204 ± 0.065||0.151 ± 0.059||3.42 ± 0.33|
| ||1.1||Active||3.46 ± 1.36||1.30 ± 2.03||0.176 ± 0.030||3.53 ± 0.16|
|2000/133||0.6||Quiet||11.52 ± 2.53||1.64 ± 0.50||0.156 ± 0.020||3.44 ± 0.10|
| ||0.6||Active||26.4 ± 9.7||5.86 ± 2.29||0.199 ± 0.043||3.66 ± 0.22|
|2000/134||0.6||Quiet||4.85 ± 0.74||0.81 ± 0.27||0.144 ± 0.026||3.38 ± 0.13|
| ||0.6||Active||33.4 ± 26.6||6.06 ± 5.22||0.163 ± 0.015||3.47 ± 0.07|
|2000/135||1.1||Quiet||0.84 ± 0.33||0.151 ± 0.076||0.168 ± 0.050||3.39 ± 0.73|
| ||1.1||Active||2.85 ± 1.41||0.54 ± 0.31||0.186 ± 0.036||3.59 ± 0.19|
|2000/136||1.8||Quiet||0.40 ± 0.10||0.079 ± 0.028||0.200 ± 0.044||3.65 ± 0.23|
| ||1.8||Active||1.25 ± 1.39||0.18 ± 0.14||0.190 ± 0.036||3.61 ± 0.18|
|2000/137||2.4||All||0.092 ± 0.030||0.021 ± 0.009||0.195 ± 0.093||3.65 ± 0.49|
 The spectral broadening data were also used to estimate electron number density fluctuation from the models. For the case of a significant flare event on 2000/134 with a measurement of BX as high as 107 Hz at 2.85 RS, the electron number density fluctuation was found to be about 62400/cm3 using appropriate adjustments to parameters in equation (8). This active period level of electron number density fluctuation is about 13 times higher than the density fluctuation estimate of 5000/cm3 during the quiet background (corresponding to quiescent period measurements of BX near 3.8 Hz earlier in the pass). The above electron number density fluctuation estimate of about 62400/cm3 during this solar maximum flare event is comparable to the total electron number density based on the Muhleman and Anderson  model developed during solar minimum conditions. This model suggests total electron number density values at 2.85RS ranging from near 50000/cm3 at the pole to 105,000/cm3 off the solar equator. Thus the actual levels of electron number density during the CME event were significantly much higher than the quiescent levels. A focus of future study would be to compare these estimates with electron density and electron density fluctuation estimates extracted from inversion of SOHO LASCO white light image brightness using techniques such as described by Hayes et al. .
 Satellite X-ray data of whole Sun X-ray flux in the 0.5–4Å (XS) and 1–8Å (XL) wavelength bands (information from the Space Environment Center, Boulder, CO, National Oceanic and Atmospheric Administration (NOAA), U.S. Dept. of Commerce) were obtained from the GOES satellite ion chamber detectors for the purpose of correlation with the Cassini radio observations of broadened bandwidth. The X-ray observations can be used to provide a sensitive means of detecting the onset time of solar flares. Given that solar flares may be related to CME events, which in turn are responsible for inducing signatures in the radio data above quiet levels, the X-ray measurements were examined to infer possible correlations with the time history measurements of B. This may allow for inference of possible relationships of the radio B measurements with the X-ray onsets near the solar surface. The occurrence of transient events in the radio data were also examined for correlation with coronal mass activity in or near the signal path using SOHO LASCO C2 or C3 white light images.
 The X-ray data exhibit two types of events above background; short duration “flares” and extended duration events [Kahler et al., 1989]. The short duration events are presumably associated with flares, which are categorized as rapid intense variations in brightness, and are not necessarily associated with flux movement such as CMEs. A solar flare occurs when magnetic field energy that has been built up is released, thus emitting radiation at wavelengths from radio, optical, X-ray and gamma. Some long duration events seen in the X-ray data appear to be associated with CMEs as related signatures are also be seen in the spacecraft radio link data. The frequency of occurrence of solar transient events has been shown to increase during periods of solar maxima in the Sun's 11 year cycle when they can be detected at any latitude off of the solar disk.
 The spectral broadening time series for each individual pass at X band and Ka band will be presented in the discussion that is to follow. These observations tend to show long “quiet” periods, where B does not vary significantly from interval to interval, as well as “transient” periods, where B increases to significantly larger values and/or dampens to or near “pre-CME” quiescent levels depending upon the activity occurring within the pass.
4.2.1. May 11, 2000 (2000/132), SEP = 1.1°
 Figure 6a displays the X band and Ka band spectral broadening bandwidth measurements, BX and BKa, versus UTC time for pass 2000/132 where the SEP angle was ∼1.1°. A gradual increase above the quiescent background of B is apparent starting about 2000/132 23:48 UTC, with a rapid increase near 2000/133 00:00 UTC where BX reaches a maximum value of 6.8 Hz and BKa reaches a maximum of 1.5 Hz at about 2000/133 00:06:40 UTC. The post-CME values fall to about BX = 3.5 Hz and BKa = 0.7 Hz beginning near ∼2000/133 00:30 UTC and continuing to the end of the X band data acquisition at 2000/133 01:00 UTC (the Ka band data actually peaks afterwards at 2000/133 01:40 UTC with a value of about 9.2 Hz). These post-CME levels lie significantly above their pre-CME quiescent levels, ∼1.1 Hz for X band and ∼0.2 Hz for Ka band. The very active period of the transient event in Figure 6a is about half (∼53%) of the data acquisition period (1.9 h/3.6 h).
Figure 6. (a) X band and Ka band spectral broadening bandwidth measurements and (b) GOES 10 X-ray flux time series during concurrent period (courtesy of NOAA), where XL denotes 1–8 Angstrom flux and XS denotes 0.5–3 Angstrom flux, for 2000/132 May 11, SEP = 1.1° (4.2RS).
Download figure to PowerPoint
 The measured ratio of the spectral broadened bandwidths, BKa/BX, better agrees with the Kolmogorov (p = 11/3) derived value of BKa/BX = 0.2 during the active period, and with the flattened turbulence spectrum value for BKa/BX = 0.11 with p = 3.2 during the quiescent period, although the statistical significance is weak. The range of power law indices estimated from the simultaneous dual-band data of p = 3.2 to 3.67 is consistent with the range cited in previous studies [Armstrong and Woo, 1980].
 The X band signal power for pass 2000/132 was in or near saturation [Morabito et al., 2003], thus the measured amplitude scintillation indices could not be used to study variability above the “quiescent” background, as these relative signal strength variations do not show any rapid increase suggestive of the flaring event during this period. A significant increase in Doppler noise however was observed in the X band frequency residuals between 2000/133 00:00 UTC to 01:00 UTC [Morabito et al., 2003].
 A coincidental possibly related signature is shown in GOES 10 X-ray data for this event (Figure 6b), which could be conjectured as emanating at the launch point of the CME at or near the photosphere surface. The signature of the X-ray data (Figure 6b) appears similar to the radio broadened bandwidth data (Figure 6a), but offset earlier in time. The peak of the X-ray feature at 2000/132 21:36 UTC occurs about 2.6 h prior to the observed peak in the radio data. A possible point of origin for the associated flare is solar feature AR 8993 where “brightenings” were seen EIT 195 (EIT 195 Angstroms is SOHO Extreme Ultraviolet Telescope image of highly ionized iron (Fe XII) taken of the solar transition region and inner corona) images at about the time of the start of the X-ray flare. Such events are similar to some having associated CMEs that have been described by Feynman and Ruzmaikin . It is cautioned that the correlations with the X-ray data and the UV brightening events are speculative at this point but are interesting and motivates further investigation.
 Figure 7a displays the X/Ka broadened bandwidth signatures versus time (Figure 6a repeated), along with SOHO LASCO C2 images taken at 2000/132 21:26 UTC (Figure 7b), and at 2000/133 00:06 UTC (Figure 7c). Figures 7b and 7c show the Cassini raypath (marked with an “X”) experiencing significant and varying levels of brightness that visually appear to correlate with the different levels of broadened bandwidth in Figure 7a (lines point from Cassini raypath impact point in Figures 7b and 7c to corresponding point on time axis for broadened bandwidth shown in Figure 7a). This CME was noted to have a position angle of 186° with 141° spread and linear speed of 716 km/s (SOHO CME catalog). The velocity measured from the SOHO images (716 km/s) is based on a fit of the expanding CME features in the images. An examination of different acceleration models (using this velocity) near the solar surface with Cassini impact point geometry shows consistency with solar feature AR 8993 being the launch point of the CME. However, this is speculative and is by no means conclusive and is a focus of further study.
Figure 7. (a) X band and Ka band spectral broadening bandwidth measurement time series, (b) SOHO LASCO C2 image at 2000/132 21:26 UTC, and (c) SOHO LASCO C2 image at 2000/133 00:06 UTC for 2000/132 May 11, SEP = 1.1°. See text for additional detail.
Download figure to PowerPoint
4.2.2. May 12, 2000 (2000/133), SEP = 0.6°
 Figure 8a displays the X band and Ka band spectral broadening bandwidths, BX and BKa, versus UTC time for pass 2000/133 (May 12, 2000) where the SEP angle was 0.6° during ingress. The Cassini broadened bandwidth data are sparse between 2000/133 18:30 to 23:00 UTC due to the temporary release of the ground station to another project (from about 2000/133 21:10 to 22:40 UTC), along with “equipment problems” as reported by station personnel in real time.
Figure 8. (a) X band and Ka band spectral broadening bandwidth measurements and (b) GOES X-ray data for 2000/133 May 12, SEP = 0.6°.
Download figure to PowerPoint
 According to the CME catalog, there was a CME at 2000/133 23:06:05 UTC that was extremely fast (2604 km/sec) and appeared as a halo, emanating from all directions about the solar disk. A significant transient event in the radio data commences at about this time and also appears to have been in effect near the end of the pass, peaking near 2000/134 00:00 UTC. The bandwidth ratio, BKa/BX, measured during the concurrent data period, follows the expected value of 0.2. The broadened bandwidth feature occurring during this period was also captured in the frequency residual data in the form of the large features seen in Figure 3. The broadened bandwidth data during this period are repeated in Figure 4c to allow comparison with the Doppler features in Figure 4a and its derived electron column density changes shown in Figure 4b, as discussed earlier. Upon a close examination of the column density in Figure 4b with the X band bandwidth in Figure 4c, there appears to be a rough increase of bandwidth with column density but nothing approaching the B ∼ (column density)2 relationship expected for the case of density fluctuations dominating the bandwidth measurements. This suggests that velocity variations may play a significant role during this CME event.
 The X/Ka broadened bandwidth signatures displayed in Figure 8a were compared with SOHO LASCO C2 images taken at selected times during the pass: 2000/133 16:50 UTC, 2000/133 19:50 UTC, and 2000/134 00:06 UTC. These images showed the Cassini raypath experiencing significant flux levels of charged particles (brightnesses) visually consistent with the differing levels of broadened bandwidth at the specific image times. This result is similar to what was illustrated in Figures 7a–7c for the case of the 2000/132 event. The sequence of time lapsed images that make up the C2 movie in the CME catalog during the period depicted in Figures 8a do show significant activity along the “streamer-like” feature that intersects the Cassini raypath. It appears that a CME front transited the Cassini raypath near the impact point during the time between the images at 20:50 UTC and 21:26 UTC.
 The GOES 10 X-ray flux in Figure 8b displays an interesting extended period double-peaked event occurring between 2000/133 21:00 to 2000/134 01:00 UTC, a 4 h period, which has elapsed between the X-ray onset and the peak in the radio data (Figure 8a). An inspection of SOHO EIT 195 images during this pass do show that significant UV activity appears to have occurred during this pass at several locations on the solar disk. These locations include solar feature AR 8993, a region near the eastern limb of the Sun, and another active region north of AR 8993 at equivalent distance from the solar equator (see Figure 9). Several EIT 195 images up to the end of this radio day show interesting events similar to those described by Feynman and Ruzmaikin . Without the missing radio data (the gaps in Figure 8a), it is difficult to draw any further conclusions or correlations with the X-ray data or EIT images.
Figure 9. SOHO EIT 195 image taken at 2000/133 19:13 UTC. Regions of known activity are annotated here such as active region AR 8993 where “flaring” was in progress.
Download figure to PowerPoint
4.2.3. May 13, 2000 (2000/134), SEP = 0.6°
 Figure 10a displays the X band and Ka band spectral broadening bandwidths, BX and BKa, versus UTC time for pass 2000/134 where the SEP angle was near 0.6°, during the egress leg of the solar conjunction. The Ka band data are continuous over the data acquisition from 2000/134 13:40 UTC to 2000/135 01:30 UTC while the X band data started late at 2000/134 16:20 UTC and also suffered an outage between 2000/134 21:20 to 23:15 UTC. The Ka band data is characterized by a very long period of quiescent background (absence of solar activity), until the occurrence of a dual-peaked event (from 2000/134 22:30 to 2000/135 01:30 UTC), which accounts for 19% of the total Ka band data acquisition period. The beginning of the event was missed in the X band data due to the outage which started near 21:20 UTC and continued until just after the peak of the first event near 23:05 UTC. The second feature at 2000/135 01:25 UTC (∼25.5 h in Figure 10a) however was captured in the X band data, shortly before the pass ended. The BKa/BX ratio lies within the range of expected values from 0.1 to 0.2. The typical mean ratio of the broadened bandwidths BKa/BX = 0.15, is indicative of a turbulence spectrum power law index of 3.2. During the solar event period, the bandwidth ratio of 0.18 lay closer to the value expected for Kolmogorov turbulence.
Figure 10. (a) X band and Ka band spectral broadening bandwidth measurements and (b) GOES X-ray data for 2000/134 May 13, SEP = 0.6°.
Download figure to PowerPoint
 The X/Ka broadened bandwidth signatures shown in Figure 10a were compared with SOHO LASCO C2 images taken at selected times during the pass, such as at 2000/134 18:06 UTC, 2000/134 23:26 UTC, and 2000/135 01:26 UTC. These images show that the Cassini raypath experienced significant flux levels of charged particles each visually consistent with the different levels of broadened bandwidth in Figure 10a. The white light images taken during the pass prior to 2000/134 22:26 UTC show that the Cassini raypath was located in a quiet region consistent with the quiescent level of the spectral broadening observed during this extended period. A brightening event was captured on the SOHO LASCO C2 image at 2000/134 22:26 UTC at the Cassini signal path location, presumably related to the first feature in the spectral broadening data which starts near that time and peaks at 2000/134 23:05 UTC, about a half hour later. A SOHO LASCO C3 image taken at 2000/134 23:42 UTC suggests some activity in the Cassini/Earth raypath region. A white light image taken at 2000/135 01:26 UTC shows an enhanced brightening in the vicinity of the location of the raypath coinciding with the second peak on the spectral broadening plot (25.5 h in Figure 10a), the data of which abruptly ends shortly thereafter. The SOHO CME catalog lists an event initiating at 2000/134 23:26:05 UTC with a speed of 966 km/s at a position angle of 241° and an angular spread of 46°, in line with Cassini's signal path. Another event to the north was noted in the CME catalog at the same time with a position angle of 353 deg and speed of 715 km/sec. The radio event near 2000/135 01:20 UTC (near 25.5 h in Figure 10a) does not appear to have an associated entry in the CME catalog but a secondary burst of material in the direction of the signal impact point is evident in the LASCO images at 2000/135 01:26:51 UTC, 01:50:06, 02:06:05 and 02:26:05 UTC which possibly could be associated with the events seen in the spectral broadening data.
 Figure 10b displays GOES X-ray data with an event of significant magnitude that appears to be nearly coincident with the transient events shown in the radio data. Here we see a double humped feature between 2000/134 23:30 UTC and 2000/135 00:30 UTC with peaks at 2000/134 23:10 UTC and 2000/135 01:30 UTC. SOHO EIT 195 UV images indicate that flares had occurred at 2000/134 19:13 UTC and 2000/135 01:13 UTC, the latter coinciding with the second event in the spectral broadening data (Figure 10a). It is interesting that the X-ray and radio spectral broadening peaks show near simultaneous occurrence. Upon examination of EIT 195 images around the time of X-ray flare initiation near 22:40 UTC, activity was noted in the region of AR 8993 where material appeared to be “ejected” in the direction of the Cassini signal impact point. Significant activity was again noted here around 2000/135 01:13 UTC coinciding with the start of the second event captured in the radio data (Figure 10a). During the flare process, interesting bright and dark features such as parallel bands could be seen in the EIT 195 images similar to those described by Feynman and Ruzmaikin .
4.2.4. May 14, 2000 (2000/135), SEP = 1.1°
 The simultaneous X band and Ka band spectral broadening bandwidth time series for pass 2000/135 (SEP ∼1°) shows that a transient event appeared to have been captured at the start at about 2000/135 16:42 UTC, then dampens down to quiescent levels as the experiment progresses by ∼2000/135 20:40 UTC, a duration of four hours [Morabito et al., 2003]. The BKa/BX bandwidth ratio average is reasonably consistent with the 0.2 value expected from Kolmogorov theory. As the Ka band data acquisition started earlier, the entire time history of the transient was captured in the Ka band data as shown in Figure 11. Here we see a quiet background for B ∼ 0.2 Hz, followed by the transient event, which increases to values as high as 3 Hz, and then dampens down to near preflare levels near the end of the data acquisition period.
 The amplitude scintillation data (Figures 3k and 3l of Morabito et al. ) show some activity near 2000/135 17:00 UTC and there do appear to be higher levels of amplitude scintillation at Ka band (Figure 3l of Morabito et al. ) during the transient event period than during the quiet period, but these levels are not significantly elevated above the background as is the case with the spectral broadening bandwidth data presented here. This suggests that increased velocity fluctuations may be dominating the bandwidth data over that of spatial plasma density fluctuations.
 SOHO LASCO C2 images at selected times during the pass: 2000/135 14:18 UTC, 16:42 UTC and 17:18 UTC show the Cassini raypath being intersected by significant and differing mass levels of charged particles. These mass levels intersecting the impact point of the Cassini-Earth raypath correlate well with the observed levels of observed broadened bandwidth (Figure 11) at the selected times, showing less intensity in white light before and after the event and increased intensity during the event.
 The concurrent GOES X-ray flux time series during this pass shows an extended period event had occurred prior to and within hours of the radio event with a comparable duration of just under three hours. Two interesting short-duration X-ray events occurred near 2000/135 17:00 UTC and 19:36 UTC which do not correlate with any features in the radio data. This is consistent with the conventional expectations that long duration X-ray events are usually associated with CMEs while short-duration X-ray events are usually not, but are instead associated with sudden bursts of brightness.
 The CME catalog shows an event occurring at 2000/135 15:26:05 UTC, with a position angle of 246° and spread of 55°, consistent with the Cassini raypath impact location being in its wake. A LASCO EIT image at 2000/135 13:13 UTC does show a brightening event was in progress near the start of the X-ray flare. An EIT 195 image taken at 2000/135 15:12 UTC shows several brightening events in this and following images taken past 15:12 UTC. Again a speculative launch point, solar feature AR 8993 shows “flashing” starting in at about 15:12 UTC.
4.2.5. May 15, 2000 (2000/136), SEP = 1.8°
 Figure 12a displays the simultaneous X band and Ka band spectral broadening bandwidth time series, BX and BKa, for pass 2000/136 where the SEP angle is ∼1.8°. Also shown are selected C3 images of coronal plasma in the vicinity of the Cassini raypath (shown as an “X”) for both an active elevated period (Figure 12b) and a quiet background period (Figure 12c). The start of the X band data in Figure 12a clearly shows the tail end of a strong event decaying down to near-quiescent levels and a second feature increasing in value starting near 2000/136 23:20 UTC at the end of the pass which may been caused by a mass of charged particles flowing across the Cassini raypath as captured on a LASCO C3 white light image at 2000/136 23:18 UTC. The CME catalog lists a CME event at 2000/136 16:26:05 UTC as being a partial halo with a velocity of about 1212 km/s at a position angle of 257° that coincides with the direction of the Cassini signal impact point.
Figure 12. (a) X band and Ka band spectral broadening bandwidth measurement time series, (b) LASCO C3 white light image at 2000/136 17:42 UTC, and (c) LASCO C3 white light image at 20:42 UTC for 2000/136 May 15, SEP = 1.8°. The arrows point from the Cassini impact point (denoted by “X”) in Figures 12b and 12c to the corresponding time in the X band bandwidth time series in Figure 12a.
Download figure to PowerPoint
 An elevated level of amplitude scintillation is apparent at the start of the X band data acquisition period at 2000/136 17:20 UTC due to a solar transient event (Figure 3m of Morabito et al. ). The amplitude scintillation signature at the 1.8° SEP angle is useful since we are in the realm of weak scattering for X band (no saturation). There is also a raised level of amplitude scintillation near the end of the data acquisition period from 2000/136 22:00 UTC to 23:30 UTC (see Figure 3m of Morabito et al. ) showing both reinforcements and degradations. The spectral broadening event occurring at 2000/136 17:20 UTC in Figure 12a is evidently correlated with a huge CME cloud moving across the Cassini-Earth raypath as shown in the LASCO C3 images taken at 2000/136 17:18 UTC (Figure 1) and at 2000/136 17:42 UTC (Figure 12b). The Cassini/Earth signal raypath region was in a relatively quiet preflare state prior to 2000/136 12:18 UTC. The X band SNR time series shows increased amplitude scintillation apparent at both the start and end of the 2000/136 pass. The increased activity is evident at the start between 17:20 UTC to 18:11 UTC and at the end of the pass after 22:00 UTC. The noise in the Doppler frequency data at the end of the pass (Figure 10d of Morabito et al. ) is not as significant as it is at the start of the pass. However, the amplitude scintillation data show significant changes at the end of the pass. A possible characterization of such an event when analyzing signals emitted from behind the corona is as follows: First an extended duration X-ray flare occurs near the solar surface. Charged particles associated with this event are hurled outward later intercepting the spacecraft-Earth signal path causing both density and velocity fluctuations, where increased levels of Doppler noise (Figure 10d of Morabito et al. ) and spectral broadening (Figure 12a) were observed as is the case around 2000/136 17:30 UTC. A quiescent period follows, and then later at 2000/136 22:00 UTC to 23:30 UTC an elevated level of amplitude scintillation occurs with a much lower level of Doppler noise and spectral broadening being observed. A reasonable conclusion is that the features at the start of 2000/136 are due to a combination of velocity and density fluctuations (shock passage) while those at the end of the pass are attributed primarily to density variations (mass flux turbulence).
4.3. Comparison of Phase Scintillation and Spectral Broadening Structure Functions
 In an attempt to examine the consistency of the phase scintillation and spectral broadening measurements, techniques were adapted from Coles et al.  to generate the composite structure functions for these data types and plot them together for selected passes (see Figure 13). The structure function provides a more convenient visualization of the distribution of the fluctuations with scale size instead of the use of spectra with wave number [Coles et al., 1991]. These structure functions tend to be steep at large scale sizes (103 to 106 km) as derived from the phase scintillation measurements and tend to display flattening at smaller scale sizes, (10 to 100 km) as derived from spectral broadening measurements [Coles et al., 1991]. An inflection was noted between steep and flat regions and was found to be abrupt occurring between 100 to 300 km scale sizes [Coles et al., 1991]. These authors also noted that the inner scale size was found to be on order of a few km very near the Sun and increased with increasing RS. The analysis results of Coles et al.  support the spectral shape model of Coles and Harmon .
Figure 13. Composite structure functions at X band for selected Cassini 2000 solar conjunction passes (solid curves) and selected Cassini 2001 solar conjunction passes (dashed curves) along with the ideal (arbitrarily scaled) Kolmogorov model (dotted black curve). These were formed by combining spectral broadening (B) structure function segments (small scales) and phase scintillation (P) structure function segments (large scales). The legend provides the pass ID in year/day of year and the solar impact distance in RS for each curve.
Download figure to PowerPoint
 Several bias effects for the structure function analysis were identified by Coles et al. . For the spectral broadening, there were biases associated with finite length of data transform, sampling rate, and smearing of the spectra due to the onboard oscillator linewidth. Passes that appeared problematic due to choice of transform parameters were not included in Figure 13. The smearing due to oscillator linewidth was deemed insignificant as the spectra were dominated by charged particle broadening for the passes shown in Figure 13. The spectral broadening structure functions were truncated at the high-end scales in order to remove portions due to estimation error associated with finite integration time for applicable cases as described by Coles et al. . The phase scintillation structure function also has some bias issues to consider. The observing plane phase departs from normal geometrical optics assumptions due to diffraction. The difference or propagation bias is more important at smaller scales where diffraction effects dominate. Coles et al.  found that there were no detectable propagation biases in any of their phase structure functions. For the data displayed in Figure 13 this bias correction was neglected.
 Figure 13 thus displays the composite X band structure functions for selected passes from both the Cassini 2000 and Cassini 2001 [Morabito, 2002] solar conjunctions derived from X band spectral broadening and phase scintillation data. The Cassini 2001 solar conjunction took place in June 2001 where the maximum of solar cycle 23 was still in progress, but was characterized with fewer solar transient events than was seen during the May 2000 solar conjunction. Earlier results such as those of Coles et al.  using these techniques were referenced to S band based on dual-band S/X Voyager 1 and 2 solar conjunction experiment data. The Voyager solar superior conjunctions also occurred during solar maximum where solar latitude variations were small and transients occurred expectedly [Coles et al., 1991]. The Cassini results displayed in Figure 13 assumed the same solar wind radial velocity dependence as used by Coles et al. , in an attempt to facilitate comparison. The derived structure function time scale was thus multiplied by a constant solar wind velocity of 400 km/s to yield the scale size shown on the x axis of Figure 13. Thus, if one desires to assume a different solar wind velocity V in km/s, the x axis can be appropriately scaled by multiplying by V/400. Since the structure function (a function of scale size, D(s)), is proportional to λ2, the curves in Figure 13 can be scaled by (8430/2295)2 (an increase of 1.1 orders) to allow comparison with the 2295 MHz S band results of Coles et al. . The Cassini data results of Figure 13 show some similarities and some differences with the Voyager results as there are cases of steepening and flattening in selected regions of the curves relative to the ideal Kolmogorov curve. The turbulence spectrum appears to be Kolmogorov during active periods and flattened during quiescent periods, which is in agreement with the work of Coles and Harmon .