A High-Frequency Radio Acoustic Sounder ior Remote Measurement of Atmospheric winds and Temperature

Transcription

1 A High-Frequency Radio Acoustic Sounder ior Remote Measurement of Atmospheric winds and Temperature Abstract The Radio Acoustic Sounding System (RASS) is used to remotely measure atmospheric temperature profiles. T h e technique used for these measurements is Doppler tracking of a short, high-intensity acoustic pulse with an R F (electromagnetic) radar. By measurement of the acoustic pulse propagation speed, temperature can be calculated as a function of altitude. The Stanford University RASS operates at an acoustic frequency of 85 Hz. Because of this low frequency and the necessity of high system gain, the unit is too large for mobile applications. Our theoretical analyses show, however, that the RASS could operate at much higher acoustic frequencies and still provide data to altitudes of 1 km even during periods of moderate to strong atmospheric turbulence. These theoretical analyses have now been supported experimentally. A RASS operating with an acoustic frequency of 1 khz not only provided Doppler data to altitudes of 1 km, but it also was able to provide a measure of horizontal winds over the same range. These experimental results came from a brief effort to support our theoretical studies. Future experiments could well extend the profiling range and versatility of the highfrequency RASS. Ultimately, we hope that our work will lead to a transportable system to be used for collecting real-time data on atmospheric winds and temperatures. 1. I n t r o d u c t i o n T h e Radio Acoustic Sounding System (RASS), presently being used at Stanford University to remotely measure atmospheric temperature, operates at an acoustic frequency of 85 Hz. Because of this low frequency and the necessity of high system gain, the acoustic source is a large (8 m X 8 m) array of folded horns. These horns, of which there are nine, measure 1.8 m high and 1.6 m across at the mouths (Frankel and Peterson, 1976). Consequently, the present RASS is immobile. Recent theoretical analyses by Bhatnagar (1976) have shown that a RASS could operate at much higher frequencies. His results show that a system with an acoustic frequency of a few kilohertz could provide data over altitudes in excess of 1 km even during periods of moderate to strong atmospheric turbulence. T h e significance of these results is that through an increase in the acoustic frequency, the size of the acoustic source can be decreased without sacrificing system gain. W i t h a smaller source, it would be possible to build a mobile 1 Stanford Research Institute, Menlo Park, Calif Equipment Development Laboratory, National Weather Service, NOAA, Silver Spring, Md Michael S. Frankel, 1 Norman J. F. Chang,1 and Melvin J. Sanders, Jr. 2 RASS allowing the user to collect real-time atmospheric data at any location he might desire. T o verify the theoretical analyses described above, personnel at Stanford Research Institute ( S R I ) designed and built a C W radar to match the 1 khz acoustic source of a pure acoustic sounder (PASS) that was designed and developed by the Equipment Development Laboratory (EDL) of the National Weather Service (NWS). Through a combination of the PASS and the C W radar, a "high-frequency" (HF) RASS was implemented. T h i s system was then used in a brief experiment to determine over what range the HF-RASS could obtain data. T h i s paper presents the results obtained from the HF-RASS experiment. Our preliminary reduction of these data verifies the theoretical predictions. In fact, it appears that not only can the HF-RASS remotely monitor temperature but it can also be used to locate wind shears and measure horizontal winds. 2. Theory of RASS operation RASS measures temperature by Doppler tracking a short, high-intensity acoustic pulse with an R F (electromagnetic) radar. By measuring VS, the speed of sound (in meters per second) in the lower troposphere, RASS allows one to calculate T from the equation (assuming still air) VS = AT\ (1) where T is the absolute temperature (in kelvins) and A is a constant that is dependent on relative humidity. T h e physical principles behind the process by which the acoustic pulse is Doppler tracked by the radar are easy to understand. Suppose a one-cycle acoustic pulse is transmitted vertically into the atmosphere. T h i s wave is longitudinal and propagates as a local condensation and rarefaction of the ambient air. These density variations cause a corresponding variation in the local index of refraction of the atmosphere, which in turn causes a reflection of a small amount of the electromagnetic energy as it propagates through the acoustic pulse. T h o u g h only a small fraction of the R F energy is reflected, RASS is able to track an acoustic pulse by virtue of two important phenomena. First, the acoustic source and the radar are collocated. Consequently, the propagating acoustic wave fronts act as large spherical Vol. 58, No. 9, September 1977

2 Bulletin American Meteorological Society 929 reflectors that focus the reflected RF signal back to the receiver. Second, an acoustic pulse consisting of many cycles can be transmitted, resulting in scattering of RF energy from successive wave fronts. Furthermore, when the acoustic wavelength (X ) is made one-half the RF wavelength (X C), the energy reflected from each acoustic wave front adds coherently at the receiver, greatly increasing the return signal strength. The condition X A = \ e/2 is basically Bragg scattering, which is the same technique used in obtaining RF backscatter from the ocean (Crombie, 1955; Tyler et al., 1974). Marshall (1972) has done a comprehensive theoretical analysis on the backscatter expected in a RASS system. His results, which were derived for ideal conditions (i.e., no wind or turbulence, and \ a = X«/2), are as follows: P r = {n 2 G rg tp tg ap a[_ 1 - cos (0/2)J/R 2 } (1.38 X )10-^10 (2) SNR = {n 2 G rg tp tg ap al 1 - cos (d/2)j/(r 2 BT)} (9.9 X 10 6 )10- LE/1, (3) where Pr received power, W; SNR signal-to-noise-power ratio; n acoustic pulse length (wavelengths) ; G r radar receiver antenna gain (ratio) ; Gt radar transmitter antenna gain (ratio); Pt radar radiated power, W; G a acoustic source gain (ratio); Pa acoustic radiated power, W; e smaller of acoustic or radar beamwidths R range, m; B radar receiver bandwidth, Hz ; T radar system noise temperature, K; L excess loss of acoustic wave, db/m. The excess loss, L, is due to attenuation of sound by absorption and scattering. These equations summarize the discussion given above. That is, because of the focusing of the RF energy by the acoustic wave fronts, the received power is only a function of the sound intensity, which decreases as R- 2. Or, as shown in (2), Pr oc Rr 2 and not R~\ Further, because of the condition X«= X e/2 (i.e., Bragg scattering), PrOC n 2 and not n. Finally, P r oc 10- LJ?/10, where L is found to increase rapidly as a function of the acoustic frequency, f a (Harris, 1967, 1971). This dependence of L on f a is the primary reason why Marshall was able to detect echoes at 1.7 km from an 85 Hz sound pulse (L db/km), whereas previous efforts at 22 khz (L > 100 db/km) were limited to ranges of 33 m or less. In the case where X a ^ X c/2, Marshall's analysis shows that P r = (y){ln 2 G rg tp tg ap a{\ - cos (6/2)) 2 /R 2^ (1.38 X 10" 16 ) 10~ LR ' 10 }, (4) where and FIG. 1. Relative reflection coefficient (7) as a function of n and \ e/\ a (R = 1 km). / sin x \ / sin y \ ~1-2f M -J- \ cos ka(rl + (5) x = (k a 2k e)n\ a/2 ; y = (k a + 2k e)n\a/2 ; k e = 27r/X e; k a = 2-7r/X a; RI distance to bottom of acoustic pulse; R2 distance to top of acoustic pulse; R2 RI n\ a. The parameter 7 can be viewed as the relative reflection coefficient when X«^ X e/2, and it is plotted in Fig. 1 for n= 10, 20, and 100. As seen from this figure, the bandwidth of 7 decreases rapidly as n increases. This dependency of 7 on n implies a trade-off in system design since both maximum bandwidth for 7 and maximum n are desired (Frankel and Peterson, 1976). 3. The HF-RASS experiment We implemented the HF-RASS by designing and building a CW radar "Bragg matched" to the EDL PASS.

3 930 Vol. 58, No. 9, September 1977 TABLE 1. System parameters. Parameter PASS CW Radar Transmitter power 340 W 1.0 W 1700 W* Antenna gain 20 db 10 db Antenna beamwidth System noise figure 10 db Frequency of operation kh ~440 MHz * Prior to 2215 LT on 22 September 1976, the PASS was operated at a peak acoustic output power of 340 W; subsequently, the acoustic power was increased to 1700 W. The system parameters are given in Table 1. The radar hardware consisted of an RF power amplifier, two helix antennas (0.25 m in diameter and 1.3 m high), an RF preamplifier, and the necessary circuitry to obtain the Doppler signal. The acoustic source was a 10 X 10 array of phase- and amplitude-matched high-frequency drivers. This array is capable of delivering an acoustic output from 40 to 2000 W, using the advertised efficiency of 40% per driver and an input power (electrical) of W. The experiment was conducted at the NWS Sterling Research and Development Center, which is situated 2 km west of the north-south runway at Dulles Airport. Also located at this facility is an NWS upper air station (DCA), which graciously provided correlative data. During the experiment the system was configured as follows. For periods of time when atmospheric winds were less than 4 m/s, the radar antennas were placed diametrically opposite each other on a circle centered about the vertical axis of the acoustic array. The minimum spacing between these antennas, 6 m, was chosen in order to keep the RF feed-through (leakage) signal low enough to avoid saturation of the RF preamplifier. Because this separation is about an order of magnitude less than the minimum altitude from which data were taken, this system configuration is referred to as a monostatic RASS. For periods when winds were greater than 4 m/s, it was necessary to configure the RASS as a bistatic radar. In this configuration, the radar antennas are aligned in the wind direction, the transmit antenna upwind from the acoustic source and the receive antenna downwind from the source. This placement of the antennas is necessary because horizontal winds move the focus of the acoustic wave fronts; consequently, the scattered RF signal level for a monostatic RASS operating during windy periods is significantly below that calculated from (2). However, displacement of the antennas as indicated above makes it possible to obtain "specular" reflection from the acoustic wave front. This reflection occurs when the axis of the acoustic wave fronts bisects the line drawn between the RF antennas. Under specular reflection conditions, signals comparable in magnitude to those calculated from (2) can be obtained even though the acoustic pulse has moved in the windward direction. In the next section, we present and discuss data obtained from these two RASS configurations. However, FIG. 2. Wind shear observed by RASS (a) and a rawinsonde (b).

4 Bulletin American Meteorological Society 931 because the objective of the experiment was only to show the feasibility of receiving RASS echoes with a system operating at high acoustic frequencies, the data are presented in the form of Doppler signal versus time. We plan to expand the present system in the near future to permit the measurement of real-time temperature and wind profdes. 4. Data Figure 2a shows typical monostatic RASS data records obtained during windy conditions. The received signal amplitude peaks at an altitude of ^100 m and decreases at a rate much greater than the L/R (R, range) predicted by (2). This echo is relatively insensitive to winds and to exact matching between acoustic and RF frequencies, and it is due to scatter from the large electromagnetic refractive index perturbations caused by the acoustic pulse when it is located just above the RF antennas. Despite efforts made to extend the range of these returns by tuning the acoustic and RF frequencies, no echoes could be received beyond 200 m on 21 September The reason for this range limitation is suggested by Fig. 2b. The rawinsonde data clearly show that there was a strong wind shear at an altitude be- FIG. 3. Bistatic RASS geometry. tween ground level and 200 m. (The RASS data indicate that this wind shear was probably between 100 and 200 m.) These experimental results are in agreement with Bhatnagar's (1976) analysis, which indicates that FIG. 4. RASS echo as a function of radar receive antenna location.

5 932 Vol. 58, No. 9, September 1977 TABLE 2. Comparison of wind measurements made by RASS and by rawinsondes. RASS Rawinsonde* Wind Wind September Local Speed, Direction, Speed, Direction, 1976 Time m/s deg m/s deg f J 130J * These values are appropriate to ~ m altitude, t Measured at 1100 LT. t Average of 1100 LT and 1915 LT rawinsonde measurements. FIG. 5. RASS and rawinsonde wind data. September The RASS measurements were made over about the 6 h period from 1030 to 1618 LT, during which time the winds were reasonably constant, as shown by the rawinsonde data. Given the surveying techniques available to us at the site (tape measure and visual sighting), we feel that the agreement between the rawinsonde and RASS data is quite good. In addition to the wind data shown in Fig. 5, we took data on horizontal wind direction and velocity at an altitude of m. Typical results for 20 through horizontal winds will displace the acoustic pulse downwind, resulting in signals well below those calculated from (2), when a monostatic RASS is used. To verify that the short echo duration was in fact due to a horizontal displacement of the acoustic wave fronts, we moved the radar receive antenna downwind to the positions shown in Fig. 3. As discussed in the previous section, our analysis indicates that this bistatic configuration should compensate for the shift of the acoustic wave fronts by the wind. Data for the bistatic RASS are shown in Fig. 4. The delays (AT) are the times required for the horizontal wind to displace the acoustic pulse to a position that bisects the line drawn from the radar transmit antenna to the receive antenna (specular condition). The limited duration of the echoes at each antenna location is due to the fact that the backscattered radar signal is still "focused" by the wave fronts. Hence, an echo is received at a given location only when the wave front is swept past the specular reflection point. These data clearly show that through movement of the receive antenna, the acoustic pulse can be tracked to almost 1 km. In this figure, as in all other RASS data shown, time delays have been converted to ranges by multiplying AT by 340 m/s, the approximate speed of sound near the ground. Hence, a 1 km range implies a 3 s delay between the transmission of a pulse and the reception of an echo. By measuring the AT between the echoes arriving at the different antenna locations shown in Fig. 3, we can calculate the average velocity of the horizontal wind component as a function of altitude. This calculation has been done for the data of Fig. 4; the results are plotted in Fig. 5, along with rawinsonde data for 22 FIG. 6. Rawinsonde data for 0700 LT, 25 September 1976.

6 Bulletin American Meteorological Society September are summarized in Table 2. As can be seen, there is generally good agreement between wind speed and direction as measured by the RASS and by the rawinsonde. On 25 September, the winds decreased to the values shown in Fig. 6. During this "quiet" day, echoes were received from near ground to km using a monostatic RASS configuration. Data for this day are shown in Fig. 7. The envelope of these selected records shows a number of interesting features. First, the large echo from just above the RF antennas is evident in each frame. Second, beyond 340 m, the low-altitude echo is not received and signals characterized by a slower decay rate can be seen. At 1130LT the envelope of the Doppler signal amplitude beyond 340 m altitude decays at about a 1 /R rate. This expected decay rate (2) implies that at this time the horizontal winds were sufficiently low that the focus of the acoustic wave fronts remained directly above the RF antennas. Third, there exists a slight enhancement of the received signal at #50 m. This enhancement is due to improved matching of RF and acoustic frequencies in this region, which occurs as a result of the temperature variations with height shown in Fig. 6. As discussed by Marshall (1972), the wavelength of the acoustic signal changes as the wave fronts propagate through regions of different temperatures; hence, the wavelength matching between the two signals (see Fig. 1) is a function of altitude. Fourth, decay rates in excess of 1 /R can be seen in a number of the far-field signals. These rates are the result of gusting horizontal winds, which moved the focus of the acoustic wave fronts from above the radar antennas. Throughout the 1-week experiment, temperatures TABLE 3. Comparison of RASS derived temperatures with NWS ground level temperatures. Temperature, C September Local 1976 Time RASS Ground above the RASS were calculated by maximizing the return Doppler signal at low altitudes. This maximization was carried out by carefully matching the acoustic and RF frequencies. Temperature (in kelvins) at near ground level was then calculated by using the following formula derived from (8) of Frankel and Peterson (1976): T = Ud/fo? where f 0 is radio frequency in megahertz, and fa is Doppler frequency in hertz (approximately equal to the acoustic frequency). Typical time-averaged RASS temperatures and correlative data from the NWS are given in Table 3. RASS data were averaged to reduce errors introduced by vertical winds (North, 1974). A point of controversy regarding the performance of an HF-RASS is the ability of the system to operate in a turbulent environment. Bhatnagar's (1976) theoretical analyses indicate that turbulence would not significantly degrade the system performance over the first few kilometers. To verify this, we waited for a day char- FIG. 7. RASS data for 25 September 1976.

7 934 Vol. 58, No. 9, September 1977 acterized by significant backscatter on the PASS data records (which we assume to be correlated with smallscale turbulence). The day selected showed four distinct backscattering structures that occurred over a range between about 100 and 500 m. On this day, we measured the power in the received Doppler signal at an altitude of 300 m as a function of acoustic pulse length. Provided that the coherency between wave fronts of the acoustic pulse was maintained while the pulse propagated through the turbulent region of the atmosphere, the receiver power would be expected to increase as n 2 (n, number of acoustic cycles, as discussed in Section 2). On the other hand, if this coherency were destroyed by turbulence and/or winds, the received power would increase at a rate <n 2. The results of this measurement are shown in Fig. 8. These data support our assertion that the coherency between the acoustic wave fronts can be maintained (even for long acoustic pulses) to ranges of at least 300 m during periods of moderate atmospheric turbulence. In the future, we intend to measure the coherency in and between acoustic wave fronts of the pulse at altitudes near 1 km. This was not accomplished during this preliminary experiment, owing to the lack of time and the need for more than two RF antennas. 5. Summary and conclusions Three types of measurements were made with the Sterling HF-RASS during the 1-week period of the experiment. First, data were collected during both windy and quiet conditions to determine the effects of winds on the maximum range of the system. We found that under quiet conditions (winds less than a few meters per second), continuous echoes could be received to km using a monostatic RASS. When winds exceeded a few meters per second and a monostatic RASS was used, radar Doppler signals from the acoustic pulse were limited to ^250 m altitude by wind shear near the ground. However, by orienting the radar antennas in the direction of the wind, and by moving the receive antenna downwind, the range to which echoes were received was increased to 1 km. In addition to increasing the range over which Doppler signals could be obtained, we found that the bistatic RASS permitted us to determine horizontal wind speed and direction. These data were obtained by measuring the location of the radar antennas and by measuring the delay times between echoes received at each antenna location. Second, ground level temperature measurements were made with the RASS by Bragg matching the RF and acoustic frequencies at low altitudes. These RASS temperatures were compared with ground measurements made by the NWS, and good agreement was found. Third, a measurement was made to help resolve the controversy regarding the possible degradation of the acoustic wave front coherency as the pulse propagates through regions of atmospheric turbulence. Preliminary results tend to support our theoretical analysis that turbulence does not significantly degrade the perform- FIG. 8. Test of acoustic wave front coherency. ance of a high-frequency RASS over at least the first 300 m, even when very long acoustic pulses are transmitted. Based on these experimental results, we feel that the HF-RASS could well prove to be a valuable tool for the remote, real-time measurement of atmospheric temperature, winds, and wind shears. Through further research, we hope to provide a mobile system capable of providing these data at any location desired by a user. Acknowledgments. The authors wish to express their gratitude to Jim Cunningham (Director of EDL) for supporting this effort and to Hans Jensen, Ray Price, and Ray Saenz for their help in setting up the experimental equipment at Sterling Research and Development Center. This program was supported by the NWS under SRI contract and by I R&D funds from SRI. References Bhatnagar, N., 1976: Interaction of electromagnetic and acoustic waves in a stochastic atmosphere. Ph.D. thesis, Stanford Univ., Stanford, Calif. Crombie, D. D., 1955: Doppler spectrum of sea echo at Mc/s. Nature, 175, Frankel, M. S., and A. M. Peterson, 1976: Remote temperature profiling in the lower troposphere. Radio Sci., 11, Harris, C. M., 1967: Absorption of sound in air versus humidity and temperature. NASA CR-647, Washington, D.C., 1971: Effects of humidity on the velocity of sound in air. J. Acoust. Soc. Amer., 49, Marshall, J. M., 1972: A radio acoustic sounding system for the remote measurement of atmospheric parameters. Sci. Rep. No. 39, SU-SEL , Stanford Electronics Labs., Stanford, Calif. North, E. M., Jr., 1974: A radio acoustic sounding system for remote measurement of atmospheric temperature. Final report, SU-SEL , Stanford Electronics Labs., Stanford, Calif. Tyler, G. L., C. C. Teague, R. H. Stewart, A. M. Peterson, W. H. Munk, and J. W. Joy, 1974: Wave directional spectra from synthetic aperture observations of radio scatter. Deep Sea Res., 21,

8 MDA OFFERS METEOROLOGICAL SATELLITE DATA SYSTEMS The unique design concepts developed for our LANDSAT ground station have been applied to produce an economical meteorological satellite system. MDA's Weather Image Processing System (WIPS) is a microprocessor based image processing and recording system which features: Reception of high resolution data from polar orbiting satellites such as VHRR & AVHRR and geostationary satellites such as GMS & METEOSAT Real-time production of computer tape and hard copy imagery on several types of image recorders Extensive radiometric enhancement capability Operator's terminal together with powerful interactive command language combines ease of operation with extensive processing capability Geometric correction Electronic enlargement of X2, X4 and X8 Annotation consists of unenhanced and enhanced gray scales,image identification and operator entered notation Longitude and latitude references Extensive Self-Test capability Complete turnkey stations are available including suitable antenna and receiving equipment, installation and training. For further information contact MACDONALD DETTWILER & ASSOCIATES LTD SHELLBRIDGE WAY, RICHMOND, B.C. V6X 2Z9 CANADA (604) TELEX: