To the best of our knowledge, we report the first determination of ionospheric distortion, comprising the simultaneous characterization of both multipath and Doppler, on wideband, transionospheric VHF (158 MHz) and UHF (422 MHz) signals. The measurements took place as part of the test phase of the United Kingdom–United States Wideband Ionospheric Distortion Experiment during the evening (∼1000 UT) of 18 January 2005. This characterization has been achieved using the ALTAIR radar at the Ronald Reagan Ballistic Missile Defense Test Site on Kwajalein Atoll (9.395°N, 167.469°E (12.87°N, 237.16°E corrected geomagnetic)) in the Pacific, in conjunction with a low Earth orbiting, constant radar cross-section, passive satellite (calibration sphere). During the period when the two-way S4 index was above ∼0.8 on both channels, the median coherency times were 43 and 96 ms at VHF and UHF, respectively (at 1.5σ). The corresponding median coherency bandwidths were 0.8 and 2.1 MHz.
 To a greater or lesser extent, the ionosphere affects all transionospheric radio frequency (RF) communications, surveillance [Xu et al., 2004], and navigation systems [Rogers et al., 2005] operating at frequencies below ∼2 GHz. The largest effects are seen at the lower frequencies; however, sensitive systems with long integration times can also be affected by the ionosphere with operating frequencies as high as 10 GHz. Perturbations of various scale sizes in the background electron density usually result in degradation of the system performance. Those small-scale irregularities which are embedded in the F region nighttime ionosphere are especially problematic [Groves et al., 1997] since they are common and cause both amplitude and phase scintillation of satellite signals.
 The effects of the ionosphere on the signal amplitude, the signal phase, and the signal group delay are best described by the complex, time-varying channel impulse response (CIR). Measurements of the CIR are, however, hard to make, and other less comprehensive measurements are generally made. These fail to elucidate whether the signal variations are due to an imposed time variant delay spread or a Doppler spread. To measure the delay spread, a wideband signal with a bandwidth greater than the coherency bandwidth is required.
 In the high-frequency (HF) band, a number of measurements has been conducted to describe the complex channel (see Cannon et al.  for a review), including those made using Doppler and Multipath Sounding Network (DAMSON), a pulse compression wideband sounder [Angling et al., 1998]. In contrast, measurements of the very high and ultrahigh-frequency (VHF/UHF) complex impulse response on transionospheric paths have not been reported. The Defence Nuclear Agency (DNA) Wideband Satellite [Fremouw et al., 1978] transmitted in the UHF band on a number of narrowband channels each separated by ∼10 MHz to elucidate frequency-dependent effects [Rino et al., 1981] but was unable to provide a measure of the CIR. Knepp and Houpis  used wideband radar measurements to measure the multipath characteristics but were also unable to determine the CIR. A number of theoretical studies have, however, been carried out [Knepp, 1983a, 1983b; Knepp, 1985; Knepp and Mokole, 1992; Knepp and Reinking, 1989].
 This paper reports the first results from a series of experiments known as the Wideband Ionospheric Distortion Experiment (WIDE). WIDE is designed to determine the time-varying transionospheric CIR, due to the equatorial ionosphere, of signals in the VHF and UHF bands. Key aims include characterizing the delay spread, frequency (Doppler) spread, phase perturbations, signal power, and signal-to-noise ratio (SNR).
 Even with the powers used in our experiment, each individual CIR can exhibit a low SNR. Sometimes, a better channel metric is the scattering function that, over a defined measurement period, gives a measure of the Doppler and the signal multipath—the temporal averaging involved increases the SNR. The scattering function can be determined by taking the absolute value of the Fourier transform of the impulse response over the time domain. In turn, this is related to the channel coherency function by Fourier transformations in both the Doppler and delay domains. More simply, the channel coherency time can be approximated as the reciprocal of the Doppler spread, while the coherency bandwidth can be approximated as the reciprocal of the multipath spread [Bello, 1963].
 In order to make these measurements of the wideband channel, we used 158 and 422 MHz signals transmitted from the Advanced Research Projects Agency Long-range Tracking and Instrumentation Radar (ALTAIR), a ground-based monostatic radar, at the Ronald Reagan Ballistic Missile Defense Test Site (RTS) on Kwajalein Atoll (9.395°N, 167.469°E). These signals were directed at, and reflected from, constant cross-section (aspect angle–independent) calibration spheres in low Earth orbit (LEO). The ALTAIR transmitter power is ≤6 MW, and the dish diameter is 42 m. For the measurements reported in this paper, the signal bandwidth was 7 MHz at 158 MHz and 18 MHz at 422 MHz, which in combination with the radar processing provided a range resolution of 32 and 12.5 m at VHF and UHF, respectively. (The sample spacings were 15 and 7.5 m, respectively.) Typically, the pulse repetition frequency (PRF) was 300 Hz but for operational reasons varied; where possible, changes in PRF were spaced more than 20 s apart to enable long periods of coherent integration to take place.
 Data associated with four calibration sphere overpasses were collected during January 2005 as part of the test phase of WIDE. January in the Pacific sector is known to have a low incidence of scintillation, and this is further compounded by the expected low levels of geomagnetic activity in the declining phase of the sunspot cycle. Well-established climatological models, such as the Wideband Model (WBMOD) [Secan et al., 1995], embody this knowledge, and at the 90% probability level this model predicted that the January 2005 Kwajalein S4 index would be less than 0.1 at 158 MHz, even during the more active postsunset scintillation period.
 This paper reports results based on the analysis of signals from just one pass of object 2826 at an altitude of ∼770 km. The measurements were made on 18 January 2005, at approximately 1000 UT (2100 LT) at a time when there was scintillation on the ALTAIR signals. A collocated L band GPS receiver recorded little to no scintillation, except on one channel where S4 reached ∼0.35 shortly after the pass. At the time of the pass, there was very strong magnetic activity with a Kp index of 8–. Very little scintillation was seen on the other four January 2005 measurement passes, as might be expected for the time of year.
 To range align with the trajectory, the target range was calculated from the UHF state vector at each pulse time. That true range was then converted to an apparent range using the radar refraction corrections; the latter was necessary since otherwise the scintillation would be superimposed on a slowly varying delay effect from the background ionosphere. Clearly, the accuracy of the ionospheric correction limits the accuracy of the range alignment.
Figure 1 describes the approximately north-south subsatellite track, which lay to the east of the radar, and Figure 2 describes the azimuth, elevation, and range of the pass during which the satellite was at an altitude of ∼770 km and reached a maximum elevation of ∼50°. Figures 3 and 4 describe the range-aligned range-time-intensity (RTI) variations at VHF and UHF, respectively, over a period of ∼10 min. Each point is based on an incoherent integration of 30 pulse returns, typically representing a period of 0.1 s; the color map is referenced to 0 dBsm. In order to achieve the latter, the received power was normalized by various system parameters such as transmitter power and antenna gain, and consequently, the nominal target gate is similarly colored in the VHF and UHF RTI plots. In order to better assess the impact of noise on our analyses, the SNR is also shown on these and subsequent Doppler-time-intensity (DTI) plots. As might be expected by reference to Figure 2, the SNR is lowest at either end of the pass (20 and 35 dB at VHF and UHF, respectively) and rises to a maximum close to the center of the pass. Evidently, our results are more contaminated by noise at either end of the pass, but the SNR is still sufficient to make meaningful scintillation measurements.
 At VHF (Figure 3), the RTI variations fall into a number of regimes as a consequence of both the radio path geometry, in relation to the irregularities, and the localization of those irregularities. Before addressing this, however, we note that the range spread extends 50–200 m from the nominal track at a low level. Furthermore, we note that this is symmetrically disposed in time about the position of closest approach (Figure 2). While it is possible that the data are influenced by tropospheric effects at either end, we conclude that this low-level range spreading is most likely due to noise.
 Addressing now the higher power (above a threshold of −10 dB), but lower range spread VHF signals, we see that these change as the satellite rises and sets. At moderate elevation angles, the range spread is typically ±45 m but with occasional excursions to 150 m for short periods of ∼10 s. These more significant variations continue until ∼340 s when the fading becomes much less severe, a situation which persists until 480 s. During this interval, the SNR remains high, and it would appear that the signal is passing through a regime with few embedded irregularities. Subsequently, the fading characteristics revert back to conditions similar to those seen as the satellite rose above the horizon. The UHF signals (Figure 4) similarly exhibit well-defined characteristics during various parts of the satellite track. As expected, at UHF, the variations are smaller, with no more than ±20 m of range spread.
 The corresponding center range Doppler-time-intensity (DTI) variations at VHF and UHF are shown in Figures 5 and 6, respectively. Each point corresponds to 1 s of data with a consequential resolution of 0.95 m s−1 at 158 MHz and 0.36 m s−1 at 422 MHz. The signal exhibits considerable variations in frequency. At VHF, values of ±20 Hz are not uncommon, with corresponding values at UHF of ±5 Hz.
 The wideband signal has allowed us to simultaneously elucidate the range and Doppler variations as a function of time, a unique measurement in this context. However, narrowband parameters can also be calculated from the data record. Figures 7 and 8 describe the two-way S4 index calculated over 10 s at VHF and UHF, respectively, and are consistent with the previously presented results; that is, periods of high Doppler and delay spread correspond to periods of higher S4.
4. Channel Scattering Function
 As previously described, the channel scattering function has the advantage of increasing the signal-to-noise ratio by virtue of temporal integration. Furthermore, it self-consistently analyzes all ranges and all Doppler frequencies rather than one range or frequency at a time. In this paper each scattering function was generated using a 1024-point fast Fourier transform typically corresponding to ∼3.4 s of data. Scattering functions corresponding to the disturbed period between 150 and 190 s are shown in Figures 9 and 10 for the VHF and UHF channels. Each is scaled to the panel peak value (white). Using a start time reference of 3.6 × 104 s, it is possible to refer back to the previously described RTI and DTI plots.
 The VHF scattering functions exhibit higher spreads than those at UHF. The Doppler spreads are easily measured at both frequencies, but the multipath delay spreads do not often exceed the instrument resolution at UHF (∼±21 ns) but often exceed the VHF value of ±53 ns. For example, by inspection, the VHF Doppler frequency spread is typically ∼±25 Hz, while the multi-path spread is typically ∼±100 ns. There is, however, much variation from scattering function to scattering function, and values as high as ∼±40 Hz and ∼±150 ns are seen. It is harder to scale the UHF signals by inspection, but these are treated analytically below.
 We also note that a number of the VHF scattering functions exhibit a horseshoe profile where the high delays are correlated with high Doppler shifts. This we interpret as being due to the satellite motion causing the radio path to move through a region of homogeneous turbulence.
 The coherency time (CT) and coherency bandwidth (CB) are variously defined. For simplicity, we have chosen to adopt the following [Cannon and Bradley, 2003]:
where ΔB is the Doppler-induced bandwidth and ΔT is the multipath spread. Both are taken to include 1.5σ (87% of the power equivalent to −4.34 dB for a normal distribution); this value was chosen to provide numerical stability.
 The scattering functions between 100 and 250 s have been analyzed to determine the coherency times and bandwidths during this period of strongest scintillation (two-way S4 ∼ 0.9 and S4 ∼ 0.8 at VHF and UHF, respectively). These have been calculated along the appropriate zero Doppler and zero multipath axes in order to be insensitive to the horseshoe spreading mentioned above. The scaled coherency values are summarized in Table 1, along with similar values scaled from the UHF data from the same period (Figure 10).
Table 1. Summary of Results
10% Coherency Time, ms
Median Coherency Time, ms
90% Coherency Time, ms
10% Coherency Bandwidth, MHz
Median Coherency Bandwidth, MHz
90% Coherency Bandwidth, MHz
 At VHF, the median coherency time is 43 ms, but 10% of the time the value is smaller than 24 ms, and 10% of the time it is greater than 85 ms. At UHF, the corresponding values are 96, 43, and 191 ms. The ratio of the medians is 2.3, which compares to an expected scaling of 2.7 (being the ratio of the two radar frequencies) [Knepp and Bradford, 1991].
 The median coherency bandwidth at VHF is 0.8 MHz, but 10% of the time it is 0.53 MHz or smaller. The upper decile value is also 0.8 MHz. At UHF, the coherency bandwidth has increased to a median value of 2.1 MHz, and the lower decile value is also 2.1 MHz. The upper decile value is 3.2 MHz. The experimental parameters dictate that the maximum coherency bandwidth which can be measured is 1.8 MHz at VHF and 3.2 MHz at UHF. No attempt has been made to scale the relative coherency bandwidths since evidently, the experiment has insufficient resolution at UHF.
5. Summary and Conclusions
 To the best of our knowledge, this paper reports the first measurements of ionospheric distortion, comprising the simultaneous characterization of both multipath and Doppler, on wideband, transionospheric VHF and UHF signals. The measurements took place as part of the test phase of the United Kingdom–United States Wideband Ionospheric Distortion Experiment (WIDE) during the evening (∼2100 LT) of 18 January 2005. This characterization has been achieved using the ALTAIR radar at the Ronald Reagan Ballistic Missile Defense Test Site (RTS) on Kwajalein Atoll (9.395°N, 167.469°E (12.87°N, 237.16°E corrected geomagnetic)) in the Pacific, in conjunction with a low Earth orbiting, constant radar cross-section, passive satellite (calibration sphere). During a period of the measurements when the two-way S4 index was above ∼0.8, the median coherency times were 43 and 96 ms at VHF and UHF, respectively. The corresponding median coherency bandwidths were 0.8 and 2.1 MHz. All measurements correspond to the spread at 1.5σ. Further data will provide experimental verification of theoretical models, which can in turn be used to aid the development of new wideband transionospheric systems such as those embodied in the U.S. VHF MUOS communications system.
 The Lincoln Laboratory contribution was sponsored by the U.S. Department of the Army under Air Force contract FA8721-05-C-0002. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the United States government. The UK contribution to this program has been conducted with funding from the Operating Environment domain of the United Kingdom Ministry of Defence Corporate Science and Technology program.