A large-aperture radiotelescope called the Solar Wind Imaging Facility (SWIFT) has been developed at the Toyokawa Observatory of the Solar-Terrestrial Environment Laboratory (STEL), Nagoya University. The SWIFT is dedicated to interplanetary scintillation (IPS) observations of the solar wind at 327 MHz, the same frequency as that of the existing STEL IPS system. The aim of this instrument is to improve the spatial and temporal resolutions of tomographic reconstructions from STEL IPS observations by increasing the number of usable lines of sight within a given time period. The SWIFT consists of a pair of asymmetric cylindrical parabolic reflector antennas with an aperture size of 108 m (N-S) by 19 m (E-W), and a 192-element phased array receiver system which forms a single beam steerable between 60°S and 30°N with respect to the zenith. Since the antenna beam is fixed in the local meridian, IPS observations are taken around the time of meridian transit for each source. The performance of the SWIFT has been tested using preliminary observations for strong discrete sources and diffuse galactic background.
 Interplanetary scintillation (IPS) of compact radio sources at meter to centimeter wavelengths serves as a powerful ground-based method for observing the solar wind plasma [e.g., Hewish et al., 1964; Coles, 1978]. The flow speeds of solar wind can be determined by measuring the time lags of diffraction patterns propagating between separate stations. The electron density fluctuation level ΔNe and the bulk density N, provided that N ∝ ΔNe, can be inferred from the observed strength of IPS. In addition, microscopic information on the density turbulence such as the spectral index, anisotropy, and inner scale can be deduced from the IPS power spectrum. The most noticeable point about the IPS method is that it allows us to observe the solar wind for a wide spatial range including the proximities close to the Sun and high latitudes. Those regions are difficult to measure by in situ, although their observations are essential for improving our understanding of the solar wind source and its acceleration mechanism. The global coverage of IPS observations is also useful for elucidating the three-dimensional structure and evolution of transient streams associated with coronal mass ejections (CMEs), which provide key information to predict space weather disturbances. Furthermore, high-latitude solar wind observations by IPS are valuable for investigating the solar cycle evolution of the solar wind, since a drastic change of polar fast winds is known to occur over the course of an 11 year period [Coles et al., 1980; Kojima and Kakinuma, 1990; Tokumaru et al., 2010]. In particular, some peculiar aspects of polar fast winds have been revealed from recent observations during the Cycle 23/24 minimum [McComas et al., 2008]. Since the Ulysses mission ended in 2009, in situ measurements over the poles have been no longer available. Therefore, further IPS observations are needed to clarify long-term evolution of the global features of the solar wind, which may provide important implications on the solar dynamo process.
 IPS observations to investigate the solar wind have been made since the early 1970s at the Solar-Terrestrial Environment Laboratory (STEL) of Nagoya University using the multistation system [Kojima and Kakinuma, 1990]. The solar wind speeds and scintillation levels have been basically determined on a daily basis between April and December from STEL IPS observations. The initial version of the STEL IPS system consisted of three antennas at Fuji, Sugadaira, and Toyokawa; and another at Kiso was added to the system in 1993 to enable four-station measurements [Asai et al., 1995]. Recently, the IPS antenna at Toyokawa has been upgraded to a more sensitive one designated as the Solar Wind Imaging Facility (SWIFT) [Kojima et al., 2003]. This paper provides an overview of the SWIFT and some results obtained from preliminary observations.
2. Science Driver of the SWIFT Development
 As mentioned above, IPS observations are useful for studying unsettled fundamental questions about the solar wind, such as the acceleration mechanism, three-dimensional (3-D) structure and propagation of CMEs, as well as long-term evolution of its global property. However, it should be noted that the parameters determined by IPS are line-of-sight (LOS) integration of actual values through the 3-D solar wind structure. The effect of LOS integration has been known to occasionally cause a significant bias in IPS observations. Therefore, compensation of the LOS integration effect is needed for any detailed analysis of IPS observations. We have developed a computer-assisted tomography (CAT) method to deconvolve the LOS integration of IPS observations so as to retrieve highly accurate data of solar wind parameters [Kojima et al., 1998; Asai et al., 1998; Jackson et al., 1998]. Several different versions of the IPS CAT method are available at present; for example, the corotational and time sequence tomography methods allow to produce fine-resolution Carrington maps of solar wind speeds or density fluctuations for single and multiple solar rotations, respectively [Fujiki et al., 2003, Kojima et al., 2007]. The MHD IPS tomography enables us to determine the 3-D distribution of solar wind parameters including speed, density, temperature and magnetic field [Hayashi et al., 2003]. The time-dependent tomography is used to reconstruct a dynamical evolution of global heliospheric features from IPS observations [Jackson et al., 2010]. The important point to note here is that the spatial/temporal resolution of all those CAT methods strongly depends on the number of LOS available for analysis, i.e., an increase in that number is essential for gaining an improved resolution of tomographic reconstructions from IPS observations. Since fainter radio sources are more abundant, a high-sensitivity system to observe IPS is required to enhance the number of LOS.
 The existing STEL IPS system was comprised of four stations located at Toyokawa, Fuji, Sugadaira, and Kiso in Japan. Table 1 shows the sensitivity of the radiotelescope, the effective area Ae, system temperature Tsys, and minimum detectable flux density ΔSmin, at each station. Here, the bandwidth B, postdetection integration time τ, background sky temperature Tsky, and physical temperature of the receiver (room temperature) Troom are assumed to be B = 10 MHz, τ = 100 ms, Tsky = 70 K, and Troom = 290 K, respectively. We note that all four radiotelescopes had nearly the same physical aperture area; ∼2000 m2, and difference in Ae in the table is ascribed to that in the efficiency of each antenna. As shown in the table, the radiotelescopes have a sensitivity adequate to detect a faint IPS signal whose flux is as weak as <1 J. That sensitivity level enables IPS observations of 30–40 sources in a day. It should be noted that the radiotelescope at Toyokawa had the poorest sensitivity among the four stations. Thus, the overall sensitivity of the existing STEL IPS system was limited by that of Toyokawa, and an upgrade of the Toyokawa radiotelescope was essential for improving the sensitivity of the STEL IPS system. That is the reason why a new radiotelescope (called the SWIFT) was developed at Toyokawa. By incorporating the SWIFT into the STEL multistation system, we intend to significantly increase the number of sources usable for IPS observations, and to achieve a finer resolution in the tomographic reconstruction from those IPS data. Note that the increase in the number of sources by the SWIFT results in an improvement of either the spatial or temporal resolution of tomographic reconstructions. If the time duration needed to accumulate IPS data is the same, IPS observations with the SWIFT allows us to investigate the 3-D structure of the solar wind at an improved spatial resolution that reveals the properties of stream interface regions and other small-scale features in greater detail. Provided that the spatial resolution is the same, IPS observations with the SWIFT allow to obtain tomographic reconstructions over a shorter observation period, thus enabling us to more precisely reveal transient features such as coronal mass ejections.
Table 1. System Noise Temperature Tsys, Effective Aperture Area Ae and Minimum Detectable Flux Density ΔSmin of the Radiotelescope at Each Station
 The tomographic reconstruction of the solar wind with a fine spatial and temporal resolution will be useful for studying not only the heliospheric physics but also the space physics around planets. The pick-up ion flux from Venus is found to greatly enhance during the passage of solar wind disturbances [Luhmann et al., 2007; McEnulty et al., 2010]. Similarly, the dynamics and loss of the Martian atmosphere are found to be strongly controlled by the solar wind condition [Edberg et al., 2010; Hara et al., 2011]. Hence, the solar wind is considered to play a crucial role in the atmosphere evolution of those planets, and continuous monitoring of the upstream solar wind which impacts on the planets is essential for elucidating the evolution process. However, in situ measurements of the upstream solar wind is usually unavailable for those planets. The tomographic reconstruction from IPS observations enables to determine reliable information of the solar wind condition at any locations including those of solar system planets, thus it can be a good complement to in situ monitoring. The potential of IPS measurements for application to the planetary research has been demonstrated from comparison between the tomographic reconstruction and magnetometer measurements at Mars [Jackson et al., 2007]. Implementation of the SWIFT to the STEL multistation system will lead to a better comparison between the tomographic reconstruction and planetary data.
3. System Specifications
 The specifications of the SWIFT are shown in Table 2, while the overall structure and dipole elements of the phased array are shown in Figure 1. A schematic illustration of the SWIFT configuration is presented in Figure 2. The observation frequency of the SWIFT is 327 MHz, the same as that of the existing STEL IPS system. The SWIFT is composed of a pair of asymmetric cylindrical parabolic reflectors with a physical dimension of 108 m (north-south) by 19 m (east-west), and a low-noise phased array receiver with 192 elements. The parabolic reflectors are fixed on the ground, and its cylinder axes are oriented north-south. The antenna directivity of the SWIFT is formed in the meridian plane by the phased array. Thus, the SWIFT is dedicated to meridian transit observations of radio sources.
Table 2. Specifications and Performance of the SWIFT
 The reflecting surface of the cylinders is made up of about 2000 thin (0.35 mm diameter) stainless steel wires at an interval of 3 cm, and is supported by four parabolic frames placed 25 m apart from one another in a north-south direction. The radio waves reflected by the cylinders are received by a phased array with a 120° corner reflector. The surface of the corner reflector is also made of thin stainless steel wires. The phased array consists of half-wavelength dipoles mounted along the focal lines of the parabolic cylinder. Here, it should be noted that the SWIFT has two focal lines corresponding to the east and west sides of the parabolic reflector, and 192 dipole elements are placed in each focal line at half-wave spacing. The whole length of the phased array is 88 m, and the cylinder aperture is 38 m (see Figure 2). Hence, the effective collecting area of the SWIFT is 3344 m2 × η, where η is the illumination efficiency. We have optimized the efficiency η by adjusting the apex angle of the corner reflector and the positions of the dipole and parabola focus, as we did for other IPS antennas at Fuji and Kiso [Asai et al., 1996]. The efficiency η of SWIFT estimated from observations are presented in section 4. In determining these structural parameters for the SWIFT, we have considered tolerance for radio interference by man-made noises in the suburban area. Since a small f/D (focal length to diameter) ratio of the prime focus antenna is favorable to reduce interference by radio waves arriving from off-beam angles, we have adopted a height of 7.2 m for the dipole position. In addition, a wire mesh fence with a height of approximately 8 m and a width of 50 m has been built along two sides of the parabolic cylinders facing north and south to shield the primary feed against suburban noises.
 Signals from a pair of east and west dipoles are combined by a T-shape joint and fed to a low-noise amplifier (LNA). Transmission loss reduction between the dipoles and preamplifier is crucial for achieving high sensitivity. We estimate the loss L = 0.52 dB. The LNA has an excellent noise performance using High Electron Mobility Transistors with the noise figure NF = 0.7 dB, which corresponds to the receiver noise temperature Tr = 52 K for the room temperature Troom = 290 K. The maximum gain of the LNA is G = 33 dB. The LNA gain and phase are varied by using a 5-bit attenuator with a minimum step of 0.25 dB and a 5-bit phase shifter, respectively. They are adjusted properly by the calibration system to form an antenna beam with a maximum available gain and minimum possible sidelobe levels. Here, we have used the loop method [Asai et al., 1997] for the calibration system. The phase/gain controlled outputs from 192 LNAs are joined via signal combiners into one signal to form a single antenna beam (see Figure 3). The beam of the SWIFT is steerable by a PC between 60°S and 30°N from the zenith. In order to reduce the signal loss at large oblique incident angles, delay cables are inserted in the signal combiner network. The received signal is transferred to an observation room, and further amplified by a superheterodyne receiver whose intermediate frequency (IF) and maximum bandwidth are 70 and 10 MHz, respectively. The IF signal is converted by a square law detector to a low-frequency (LF) voltage signal, which is digitized by a 16-bit A/D converter. In the digitization, a median mean method is employed; i.e., first, the LF signal is sampled at a rate of 10 kHz, and then a time series data with a sampling period of 20 ms is derived by taking the median value of the sampled data. This method eliminates impulsive noises from the sampled data more effectively than the arithmetic mean method.
 Observations with the SWIFT are fully automated by referring to schedule files which include information of radio sources, start/end times, and receiver parameters. Radio sources are selected from those used for IPS observations at STEL and Ooty (P. K. Manoharan, private communication, 2007) by considering the solar elongation ɛ and quality as a scintillating source. Our IPS observations are conducted for ɛ < 90°. The IPS data are collected around the meridian transit time of a given radio source, with the duration time for each IPS observation being limited to 2.7 min in order to minimize the decrease in the IPS signal due to off-beam observations. While IPS observations are the first priority for the SWIFT, observations of some nonscintillating sources with an intense flux are made during the night to monitor the beam efficiency. For this observation mode, we employ 10 sources whose fluxes are well known. A longer duration (usually 40 min) is used here to cover a whole beam pattern including the first sidelobe. Furthermore, a special mode for mapping observations of the radio sky is available. In this mode, the antenna beam is repeatedly swung within a given elevation range to scan the radio sky using the Earth's rotation (see section 4). Data collected with the SWIFT are transferred daily from the Toyokawa Observatory via the Internet to the STEL headquarters in Nagoya for conducting various analyses.
4. Preliminary Results
 Here we present some preliminary results from initial observations with the SWIFT.
4.1. Mapping Observations of 327 MHz Radio Sky
Figure 4a shows a radio sky map at 327 MHz observed with the SWIFT. This map is produced by swinging the antenna beam between 60°S and 30°N from the zenith; −25°S and 65°N in declination. An increase in the system temperature near the horizons has not been corrected in this map. The radio sources that can be identified from this radio map include diffuse sources distributed along our Milky Way, and discrete sources such as Cassiopeia-A (Cas-A), Cygnus-A (Cyg-A), and Taurus-A (Tau-A). Note that artificial structures caused by the SWIFT sidelobes arise in the east and west of the primary beam for some strong sources. A diffuse structure emerging from the galactic center and extending north (the so-called north spar) are clearly discernible, while other fainter diffuse structures also are observable on this map. Such features revealed by SWIFT mapping observations are consistent with those reported from earlier all-sky surveys at UHF frequencies. Figure 4b shows a all-sky map produced from 408 MHz observations [Haslam et al., 1982]. The angular resolution of this map is 0.85°, comparable to that of the SWIFT. The marked resemblance between two maps suggests that the SWIFT is capable of making excellent radio astronomical observations. More detailed discussion is deferred to a future study, since it requires careful calibration of the SWIFT data before comparing them with earlier results.
4.2. Estimation of Antenna Efficiency
Figure 5 shows variations in signal intensity observed for 3 hours around the meridian transit time of Taurus-A (3C144) on 5 March 2009. In this observation, the receiver noise level was measured at a 10 minute repetition period (vertical bars in Figure 5). The beam pattern of the SWIFT in an east-west direction is clearly revealed from this measurement, and the pattern is in good agreement with the one expected from a combined system having two uniform apertures with a length of ∼14 m and a spacing of ∼21.5 m. If the flux S of Taurus-A and the noise temperature Tant at the antenna output are given, the antenna efficiency η of the SWIFT can be calculated from the data shown in Figure 5. Here, we assume that S = 1200 J for Taurus-A at 327 MHz. The noise temperature at the antenna output is given by Tant = TskyL + Troom (1 − L), where Tsky, Troom, L are the background sky temperature, room temperature and transmission loss between the antenna and the LNA, respectively. Assuming that Tsky = 70 K, Troom = 290 K, and L = −0.52 dB, we obtain Tant = 94 K. Using this value, the signal power of Taurus-A received by the SWIFT is determined from the data of Figure 5 to be P = 1.06 × 10−13 W, which corresponds to the effective aperture area Ae = 1820 m2 for the source direction and Ae = 1970 m2 at the zenith, since the bandwidth is 10 MHz. Hence, the antenna efficiency η is estimated as η = 0.59. This value is generally consistent with the reduced aperture size (∼14 m) inferred from the beam pattern. Since the system noise temperature is Tsys = Tant + Tr = 146 K, the minimum detectable flux density of the SWIFT is estimated to be ΔSmin = 0.2 J, if the same integration time τ = 100 ms as that in the existing IPS system is used, and ΔSmin = 0.46 J for τ = 20 ms.
4.3. IPS Observations
 IPS data observed for two scintillating sources, 3C48 and 3C49, are displayed in Figures 6a and 6b, respectively. Figures 6a (top) and 6b (top) and Figures 6a (bottom) and 6b (bottom) show the time variation and power spectrum of signal intensity. Observed spectra (black lines) clearly exhibit a typical IPS feature; a flat level at low frequencies and a steep fall at higher frequencies. The transition point between the flat and steep spectra corresponds to the Fresnel frequency, which can be related to the solar wind speed [e.g., Rufenach, 1971]. It should be noticed that the spectral slopes above the Fresnel frequency are significantly different between the two observations, even though they are taken at nearly the same time. This fact is ascribed to different properties in the solar wind turbulence along the lines of sight corresponding to the two sources. Thus, the shape of the IPS power spectrum reflects the physical parameters of the solar wind, and the spectral fitting analysis has been used to extract those parameters from IPS observations [Manoharan and Ananthakrishnan, 1990; Tokumaru et al., 1992]. In Figure 6, the best fit curves of the IPS power spectrum model are also indicated by red lines. The model used here assumes weak scattering of radio waves and a power law spectrum of the solar wind turbulence. The solar wind speeds V and spectral indices α determined by this model fit are V = 499 km/s and α = 4.34 for 3C48 and V = 379 km/s and α = 3.17 for 3C49. A fuller discussion of the result is beyond the scope of this paper, and will be presented in a separate paper.
 We have developed the SWIFT, which is a radiotelescope dedicated to IPS observations of the solar wind at 327 MHz. The SWIFT exhibits an excellent sensitivity enabling us to observe the IPS for radio sources weaker than those we used previously. The performance of the SWIFT has been demonstrated from preliminary observations of discrete sources and diffuse background. IPS observations with the SWIFT have been carried out continuously since 2008 summer, and data of the IPS power spectrum and strength are available from those observations. We note that a further development is required to determine the solar wind speed by taking cross correlation between the SWIFT and other STEL IPS radiotelescopes. Hence, a project to upgrade the existing IPS radiotelescopes at Fuji and Kiso is currently in progress. This project aims to determine the solar wind speed from simultaneous IPS observations with the SWIFT and those telescopes.
 When multistation IPS observations including the SWIFT are available, they will provide an important opportunity to carry out the following studies; First, the IPS data will enable us to reveal global features and the evolution of solar wind in the solar cycle 24. Peculiar aspects of the solar wind in the cycle 23/24 minimum have been reported from several studies including our IPS observations [Tokumaru et al., 2009, 2010]. Such peculiarities are attributed to the weak polar field in this cycle. Since the evolution of the solar wind in the cycle 24 is likely to significantly differ from that in previous cycles, our IPS observations of the cycle 24 may be useful in gaining insight into the physical processes of the solar dynamo and wind formation. Second, our IPS observations will provide comparison between solar wind data obtained with the multistation method and those obtained with the single-station method (i.e., the spectral model fit). Such a comparison is essential to examine a possible systematic bias between them and thus to improve the accuracy of IPS observations.
 We expect that multistation IPS observations using the SWIFT will be available in 2011. However, further efforts to maintain and improve the sensitivity of the multistation IPS system are needed to accomplish the studies mentioned above. The installation of LNAs with a lower noise temperature for Fuji and Kiso telescopes is regarded as one of the most essential tasks for improving the overall system sensitivity.
 This work was supported partly by the “Survey of Energy Transfer Process in Geospace” a special fund for education and research, and partly by Grants-in-Aid for Creative Scientific Research (17GS0208) and Scientific Research (B) (21340140) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. The IPS observations were carried out under the solar wind program of the Solar-Terrestrial Environment Laboratory (STEL) of Nagoya University.