ELECTRIC WIRE & CABLE, ENERGY

Development of Tropospheric Wind Profiler with Luneberg (WPR LQ-7)

Katsuyuki IMAI, Takao NAKAGAWA and Hiroyuki HASHIGUCHI

Sumitomo Electric Industries, Ltd., in joint research with Kyoto University's Research Institute for Sustainable Humanosphere, has developed "LQ Series", a new model of wind profiler radar system, for measuring the velocity and direction of winds using radio waves. This new model, which employs a antenna, has substantially lower price, improved characteristics and increased user-friendliness than previous models. In this paper, the authors explain technical information of the new model wind profiler (WPR LQ-7).

1. Introduction observed height, an extension of system life, and an improvement in maintenance ease. The atmosphere that covers the earth is roughly This paper reports on the technical aspect of WPR divided into four layers: starting from the earth’s surface LQ-7 (Photo 1) and the observation result achieved by is troposphere, stratosphere, mesosphere and thermos- this new model. phere. Especially the troposphere (up to about 10 km) is the region in which the atmospheric motion is very complicated, due to friction with the earth's surface, heat radiation from the ground and heat absorption into the ground. As a result of complicated atmospheric motion, various weather changes are caused, such as localized torrential downpour, wind sheer, and so on. The weather observation instruments that can observe the air motion of the troposphere with high degrees of accuracy and time resolution has been developed in var- ious countries, especially in European countries, the US and Japan, due to the remarkable progress of the remote sensing technology in recent years. Among such instruments, wind profiler radar (WPR), which can pro- vide the three-dimensional structures of wind velocities and wind directions of each layer of atmosphere in real time, has become indispensable to the world today. Sumitomo Electric Industries, Ltd. and the Radio Photo 1. WPR LQ-7 Science Center for Space & Atmosphere (RASC) (present- ly Research Institute for Sustainable Humanosphere (RISH)) at Kyoto University have jointly started to devel- op L-band radar for boundary layer observation (up to 3 2. Principle of WPR km) in 1994, and succeeded in the development of the world’s first transportable S-band boundary layer radar Figure 1 schematically shows the principle of opera- in 1998. In 2002, Sumitomo Electric developed a tropos- tion of WPR. Electromagnetic pulse waves radiated by pheric WPR (model number: WPR L-28) whose observa- the radar antenna propagate toward the sky. During the tion height was drastically improved. These WPR sys- propagation, the electromagnetic pulse waves experi- tems are being installed in Japan, China, South Korea, ence random refractivity fluctuations caused by atmos- and used for observations. pheric turbulence and are scattered. Parts of the scat- Ever since Sumitomo Electric’s WPR products tered pulse waves (echoes) then return to the radar with became widely used, demands for higher observed time delays proportional to the height where the scatter- height and lower price have been rising. Sumitomo ing had occurred, making it possible to relate the scatter- Electric decided to make the full model change of its ing intensity to the height by sampling them with proper existing WPR L-28 model to meet these demands. The time intervals. Since turbulence moves with the flow of Company has developed new model WPR LQ-7, which the wind, the echoes are subjected to the adopts Luneberg lens in the antenna part by using its shifts (Doppler shifts) proportional to the wind velocity advanced material technology. Adopting Luneberg lens (V ) at the height where the scattering took place. as an antenna resulted into a big reduction in cost (40% The following equation holds between the Doppler reduction from the previous model), an expansion of shift ∆f and the radial wind velocity Vr.

38 · Development of Tropospheric Wind Profiler Radar with Luneberg Lens Antenna (WPR LQ-7) c Vr at the same phase, the S/N of the scattering wave can be ＋ ...... ∆f ＝f 0 －1 (2-1) improved without sacrificing height resolution. c －Vr Here, f0 is the radar frequency and c is the velocity of light. Furthermore, since Vr is negligible compared with c, Equation (2-1) can be approximated to Equation (2-2). Height

∆ f hi+N-1 Vr ＝c ...... (2-2) ti = t1 + (i – 1) τ 2f 0 c 1 hi = — { ti – (N – —) } τ hi 2 2 Hence, when the beam radiated from WPR is hN (i = 1,2,···) directed to the zenith, the vertical component of the h1 wind velocity Vz is determined from Equation (2-2). By θ tilting the beam ± degrees in either direction from C1 C2 CN zenith and measuring the radial wind velocity Vr(θ), the τ t1 t i-N+1 t i Time horizontal component of the wind velocity Vh is given τ by Equation (2-3), on the assumption that the wind N within the beam scanning field is uniform. Vr（θ）－Vr（－θ） � Vh＝ ...... (2-3) 2 sinθ� Decoding Transmitted Pulse π π Based on the principle described above, the distrib- Encoding Phase 0 Compression (Phase Modulation) 1 -1 1 utions of the wind’s direction and velocity with height Code are obtained. Direct observation of the vertical compo- N times nents of wind velocity, which has not been achieved by other observation instruments, is one of the striking fea- tures of WPR. Resolution τ Original Pulse

Fig. 2. Principle of pulse compression technique Height hm Vh hn+1 hn hn-1 h1 V Vz In addition, if the complementary code sequences h1 tn-1 tn tn+1 tm Time for optimum decoding, which are devised by E. Spano f Vr and O. Ghebrebrhan [1996] (1), are adopted, it becomes -θ f+∆f θ possible to perform the pulse compression in the low Transmitted Pulse altitude, and the side-lobe and the influence of the θ - θ S1 Sn-1 Sn Sn+1 Sm interference wave are suppressed.

Sampling The new system (WPR LQ-7) employs this optimum S1 Sn Sm complementary code sequences. h1 hn hm

c × tn hn c Light Velocity ＝ 2 （ ： ）

Fig. 1. Principle of WPR 3. Principle of Luneberg Lens

The principle of Luneberg lens is shown in the Fig. 3. Since the scattered waves that serve as the basis of Luneberg lens is the lens which R. K. wind velocity detection are extremely feeble and their Luneberg developed in 1944. A relative permittivity levels of scattering become lower as the scattering changes corresponding to the distance from the center height becomes higher, the pulse compression tech- of the spherical dielectric, and the incident plane wave nique was implemented in order to improve S/N. is gathered into one focal point on the surface symmet- The principle of pulse compression is shown in Fig. 2. rical to the center of the dielectric. Because the shape of The received voltage at the sampling time ti is the sum of the dielectric is spherical, each of the arrived electro- scattering waves from the height in the range of hi to magnetic waves from every direction has its own focal hi+N-1. When the phase of each transmitted pulse is modu- point. Therefore all of the points on the lens surface lated at intervals of time τ, scattering wave from each can become focal points, and the electromagnetic waves height mentioned above shall be multiplied by the trans- from arbitrary directions can be received independent- mitted pulse’s phase coefficient Cp (p = 0, 1, ····, N-1). ly. Conversely, the electromagnetic wave radiated from When the scattering wave at each height is calculat- the focal point is radiated as a plane wave after traveling ed by combining the received voltage at ti-p+1, because through the lens. These characteristics suggest that no correlation exists among the phases of the noise com- Luneberg lens works as a multi-beam transmission and ponents, etc., whereas the scattering waves are added up reception antenna.

SEI TECHNICAL REVIEW · NUMBER 64 · APRIL 2007 · 39 outodoor unit Indoor unit Because the Luneberg lens has data an infinite number of focal points, it allows the electromagnetic Signal Processing Unit Electromagnetic Luneberg Lens waves from arbitrary directions control Antenna unit Wave to be focused independently and simultaneously. Transmitter unit power power control Focal Point UPS data

Power Unit Data Processing Unit 2.0 1.8 1.6 WAN 1.4 2 Router r Remote Control Unit 1.2 ε r = 2 – 1.0 R Dielectric Constant 0 0.2 0.4 0.6 0.8 1 ε r : Dielectric Constant Fig. 4. System configuration Center Surface r : Radius Relative Radius R : Lens Radius

Fig. 3. Principle of Luneberg lens Data Processing Unit PC (Windows XP)

Luneberg Ether 100base-T 4. System Outline Lenz 1357.5MHz HPA Control CPU Status Feed DSP selector T/R The main specifications of WPR LQ-7 are shown in Radar controller

Table 1, and the configuration of this system and its LNA Pulse T/R T/R block diagram are shown in Fig. 4 and Fig. 5, respective- modulation ly. The system is composed of seven antenna units, a IQ Local IF transmitter unit, a power supply unit, a signal processing 1227.5MHz 130MHz unit, and a data processing unit (PC). Antenna Unit Transmitter unit Signal Processing Unit

Table 1. Specifications of WPR LQ-7 Fig. 5. Block diagram Operational Frequency 1.3575GHz / 1.290GHz

Antenna Active Phased Array 4-1 Antenna Unit An antenna unit is equipped with an 800-mm-diam- Antenna Gain >30dBi eter lens antenna whose gain is about 21 dBi. Seven (Az, Ze) = (0˚, 0˚), (0˚, 14˚), antenna units are arranged 900 mm apart, and the Beam Directions (90˚, 14˚), (180˚, 14˚), (270˚, 14˚) antenna gain of 30 dBi is achieved in a total. The num- ber of antenna units in this system can be changed Polarization Linear freely according to the desired observed altitude, Peak Power >2000W adding expandability to the system. Examples of typical arrangements of antenna units (LQ-4, LQ-7 and LQ-13) Average Power >700W are shown in Fig. 6. The antenna pattern of LQ-7 is shown in Fig. 7. 333ns, 666ns, 1000ns, 1333ns, Pulse Length 2000ns, 2666ns, 4000ns Under the lens, five primary feeds (for five beam directions: zenith, north, south, east and west) are IPP 50, 80, 100, 120, 150, 200µs arranged at a zenith angle of 14 degrees. High-speed beam scanning is realized by electronically switching the (1), 2, 4, 8, 16 bits (Optimum Pulse Compression primary feed that transmits and receives an electromag- Complementary codes)

Noise Figure <2.0dB

Dynamic Range >60dB [WPR LQ-4] [WPR LQ-7] [WPR LQ-13]

Z Y Z Y Z Y Coherent Integration Variable (<200)

FFT Points 64, 128, 256 (default), 512

Power Supply 1ø-200V X X X 4000mm 4000mm 4000mm Operational Temp. -30˚C - 50˚C (Outdoor unit)

Wind Durability >90m/s (moment) Fig. 6. Typical arrangements of antenna units

40 · Development of Tropospheric Wind Profiler Radar with Luneberg Lens Antenna (WPR LQ-7) LQ-7 Antenna Pattern (1.3575GHz) The signal processing part is composed of a PC 30 (Windows® XP), to which an A/D-DSP board and a 25 radar controller board are connected by a PCI-bus. The 20 signal processing flow is shown in Fig. 8. The received 15 signal is digitized with an A/D converter (2ch, 14 bit, 3 MHz). After real-time signal processing such as pulse 10 compression decoding, coherent integration, FFT and Gain [dBi] 5 incoherent integration, a raw data (spectrum data) is 0 generated and then stored in the hard disk. The raw -5 data (Raw) is transmitted through LAN to the data pro- -10 cessing unit at least every one minute. -180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180 angle [deg]

Tipp Fig. 7. Antenna pattern of WPR LQ-7 Process the number of data points 12 NRData Sampling Decoding of netic wave. The improvement of high-speed beam scan- 1 2 Nptn Pulse compression NR ning capability by shortening of switching cycles 12 NcCoherent Integration 12 NFFFT improves the simultaneity in data acquisition between NR * NF each beam direction. Furthermore, the use of data from 12 NBBeam scanning opposing two beams allows precise measurement of ver- 1 2 Ninc Incoherent Integration NR * NF * NB tical flux of horizontal wind moment, which is of impor- 1 Unit observation data tance in meteorological studies.

Each antenna unit is respectively equipped with a Tmeasure Total data size = 4bites * NR * NF * NB [bites/min.] receiver of a noise figure of 1.5 dB, and suppresses the feeder loss by equipping it under antenna unit. 4-2 Transmitter Unit Fig. 8. Signal processing flow The transmitter unit is equipped with seven pairs of frequency converter and transmitter. The active phased array system was adopted to use several transmitters 4-4 Data Processing Unit instead of one and make the output power of each trans- The data processing unit is a PC (Windows® XP) . mitter small (peak power: 400 W, duty ratio: 35%). The Firstly, the nonlinear least square fitting by polynomial effective total output power of the antenna is 2000 W or function is carried out to the Raw data, which is sent greater (peak power). The transmitters adopt the GaN from the signal processing unit. The basic elements power devices manufactured by Eudyna Devices Inc. (a (peak value P0, Doppler shift fd and spectrum width σ) related company of Sumitomo Electric), successfully for evaluating wind direction and velocity, that are cal- achieving the drastic improvement of power consump- culated by the fitting described above, are stored as tion compared to the case using GaAs power devices. moment data (Moment) in the hard disk. Various control signals that allow active operations Furthermore, data quality control (error determina- of antenna units and transmitter unit are generated in a tion) is carried out on the Moment data. First, abnormal radar controller, which is involved in the signal process- data in each element of the Moment data are detected ing unit to be mentioned later. The timings of genera- and deleted. Next, data of ten minutes are collected and tion of these control signals are synchronized with the the continuity in time-space is verified. If abnormal data is A/D sampling timings. detected, it is deleted from each Moment data. Finally, the Operation statuses such as transmission frequency, ten-minute quality controlled Moment data is averaged transmitter output power, DC power failure, abnormal and stored as average data (Average) in the hard disk. internal temperature and cooling fan failure are detected The data processing unit can display a real-time as radar status signals, and are transmitted to the signal observation data on the screen. The following is the processing unit so that remote monitoring is possible. data actually observed by WPR LQ-7. Figure 9 is the 4-3 Signal Processing Unit example of simultaneous display of Raw data obtained The signal processing unit is composed of a signal by five beams. The abscissa axis indicates radial wind processing part, which performs data sampling and sig- velocity: negative wind velocity represents the direction nal processing, and an analog part, which is composed receding from the radar and positive wind velocity rep- of an IF oscillator (130 MHz) and a baseband system. In resents the direction approaching to the radar. The the analog part, the IF oscillator’s transmitter and mod- ordinate axis indicates echo intensity (reception intensi- ulator generate an encode pulse (BPSK: 16 bit) for the ty) and the depth direction indicates the height. This previously-described pulse compression and transmit it figure shows that wind velocity changes with height. to the transmitter unit. Moreover, a receiving signal is Figure 10 shows the time-height profile of horizon- detected with an IQ orthogonal detector and transmit- tal wind which is indicated by wind barbs, with the echo ted to the A/D converter. intensity (receiving intensity) indicated in colors super-

SEI TECHNICAL REVIEW · NUMBER 64 · APRIL 2007 · 41 imposed on it. Each wind barb represents the direction of wind in the horizontal plane: the upward direction in the diagram shows northward wind and the rightward direction in the diagram shows eastward wind. This soft- ware allows superimposition of parameters other than echo intensity, such as vertical wind velocity, S/N and Cn2, as well. 5. Conclusion

The tropospheric wind profiler radar (WPR LQ-7) that adopts Luneberg lens in its antenna was successfully developed. Observation data in clear atmosphere has achieved echo detection of the height 7 km or higher, which proves that WPR LQ-7 is useful for the observa- tion of troposphere.

Fig. 9. Example of simultaneous display of Raw data obtained by 5 beams

References (1) E.Spano and O.Ghebrebrhan , Sequences of complementary codes for the optimum decoding of truncated range and high sidelobe suppression factors for ST/MST radar systems, IEEE Transactions on Geoscience and Remote Sensing., 34,330-345,1996.

Fig. 10. Time-height profile of horizontal wind (Echo intensity is indicated in colors.)

Contributors (The lead author is indicated by an asterisk (*)). K. IMAI* • SEI Hybrid Products, Inc. T. NAKAGAWA • Sumitomo Electric System Solutions Co., Ltd. H. HASHIGUCHI • Ph.D., Associate Professor, Research Institute for Sustainable Humanosphere (RISH), Kyoto University

42 · Development of Tropospheric Wind Profiler Radar with Luneberg Lens Antenna (WPR LQ-7)