. 1, *S5\ & SZ3 : 52, znJS*™ NQ^/EI^NA^vePropagationl^boratiory National Genter for Atrnospheric Research THE BOULDER LOWLEVE! INTERCOMPARISON (I <*> EXPERIMENT Preprint of WMO Report Digitized by the Internet Archive in 2012 with funding from LYRASIS Members and Sloan Foundation http://archive.org/details/boulderlowleveliOOkaim THE BOULDER LOW-LEVEL INTERCOMPARISON EXPERIMENT Preprint of WMO Report Editors: J. C. Kaimal H. W. Baynton J. E. Gaynor Report Number Two June 1980 NOAA/NCAR Boulder Atmospheric Observatory A joint publication of NCAR and NOAA available from NOAA/ERL, Boulder, CO 80303, and from the NCAR Publications Office, P.O. Box 3000, Boulder, CO 80307. NOTICE Mention of a commercial company or product does not con- stitute an endorsement by NOAA Environmental Research Laboratories. Use for publicity or advertising purposes of information from this publication concerning proprie- tary products or the tests of such products is not au- thorized. This report is to be published in a series of the World Meteorological Organization . For sale by the Superintendent of Documents, US. Government Printing Office, Washington, D.C. 20402 (Order by SD Stock No. 003-017-00480-5) FOREWORD (to be added by WMO) PREFACE In preparing this report for the WMO Commission for Instruments and Methods of Observation (CIMO), the editors have aimed at providing adequate coverage of all three aspects of the Boulder Low-Level Intercomparison Experi- ment (BLIE): the workshop preceding the experiment, the experiment itself, and the results of the intercomparison. Papers presented at the workshop form the first part of the report, Chapters 1-23. The description of the experi- ment in Chapter 24 reflects the perspective of the three members of the orga- nizing committee who planned and directed the experiment. Chapter 25, pre- pared by the same three authors, describes the results and includes comments from participants on the performance of their sensors and explanations for possible discrepancies in their results. Here we have tried to remain faith- ful to the spirit of the discussion session held at the end of the experiment, when preliminary results from the experiment were first examined and discussed. Our objective is to improve the present understanding of low-level sounding techniques, not to rank sensors according to performance. The experiment was conducted in August and September 1979 at the Boulder Atmospheric Observatory (BAO) , located 25 km east of Boulder, Colorado. It was divided into the following phases: (1) Checkout and preparation, 20-22 August. (2) Workshop, 23-24 August. (3) Comparison tests, 27 August-5 September. (4) Discussion session, 6 September. Scientists from the National Oceanic and Atmospheric Administration (NOAA) of the U.S. Department of Commerce and from the National Center for Atmospheric Research (NCAR) collaborated in the planning and direction of all phases of the experiment. The BLIE organizing committee consisted of the following members: J. C. Kaimai, Chairman (NOAA), H. W. Baynton (NCAR), J. E. Gaynor (NOAA), F. G. Finger (NOAA), W. U. Weimann (WMO/CIMO) , and H. P. Treussart (WMO/CIMO). Readers of this report will note that three of the measuring tech- niques described at the workshop were not tested during the experiment. The descriptions of those techniques contain data from an experiment conducted at the same site a year earlier under very similar conditions (Project PHOENIX, September 1978). The instrumented aircraft, dual-Doppler radar, and passive radiometry techniques required such extensive data processing that guidelines imposed on other participants for data submission could not be applied to them. Nevertheless, those techniques were considered important enough to be included in the workshop. Participation in BLIE required that processed data be submitted within 24 hours of any operating period or that analog voltages be made avail- able for real-time sampling and processing on the BAO data-acquisition systems. Without these restrictions the discussion session held after the experiment to evaluate the results would not have been possible. Years of delay, disagree- ments, and frustrations over the outcome can be expected when participants do their processing after returning home from the experiment. Eleven member nations of WMO participated in the experiment with equipment or as observers. Both the visitors and the resident staff of NOAA and NCAR cooperated to make this complex operation a success. Many of them put in long hard hours in the field. We take this opportunity to express our gratitude to all the participants. We are happy to acknowledge the financial and moral support provided by NOAA management: Dr. Wilmot N. Hess, Director, Environmental Research Laboratories, and Dr. C. Gordon Little, Director, Wave Propagation Laboratory. Facilities for holding the workshop and discussion sessions were generously provided by NCAR management: Dr. Francis P. Bretherton, President, UCAR, and Dr. John W. Firor, Executive Director, NCAR. We are also grateful to WMO for financial assistance extended to participants who would otherwise have been unable to attend, and to the participating or- ganizations and member nations for their strong endorsement of this compar- ison. An experiment of this magnitude would not have been possible without the full support and commitment of resources that accompany such endorsement. J. C. Kaimal H. W. Baynton J. E. Gaynor Editors CONTENTS Page Foreword (to be added by WMO) iii Preface v Part 1 PROCEEDINGS OF THE BLIE WORKSHOP 1. BAO Sensors for Wind, Temperature, and Humidity Profiling J. C. Kaimal 1 2. Single-Head Sonic Anemometer-Thermometer T. Hanafusa, Y. Kobori, and Y. Mitsuta 7 3. Vaisala Eight-Level Instrumented Tower System I. Ikonen 14 4. Remotely Piloted Aircraft for Atmospheric Soundings (SAM) D. Martin 18 5. Comparison of Aircraft and BAO Tower Measurements D. H. Lenschow and B. B. Stankov 26 6. Tethered Aerodynamical ly Lifting Anemometer (TALA) C. F. Woodhouse 33 7. Remote Acoustic Electronic Sounding (RACES) P. Ravussin 38 8. FM-CW Radar R. B. Chadwick and K. P. Moran 47 9. Dual-Doppler Radar R. A. Kropfli 50 10. Remote Sensing of Temperature Profiles with Combined Active and Passive Sensors M. T. Decker 59 11. WPL Doppler Sounder W. D. Neff, H. E. Ramm, and C. Wendt 63 12. Doppler Acoustic System for Wind Profiling (AVIT) P. MacCready 70 13. Radian Corporation Model 800 Echosonde M. A. McAnally 77 14. Swedish Sodar System S. Salomonsson and M. Hurtig 81 15. The X0NDAR R. L. Peace, Jr 87 16. GMD-1 Radio Wind Sounding System R. B. McBeth 98 vii 17. TDFS Low-Level Radiosonde System C. Fink, E. Schollmann, and A. Kb'lbl 100 18. CORA Radiosonde System Using Free-Flying Balloons I. Ikonen 105 19. The Airsonde System D. B. Call and A. L. Morris 108 20. The Tethersonde System A. L. Morris and D. B. Call 117 21. Tethered Balloon Profiler System K. Stefanicki 125 22. Boundary Layer Packages for Tethered Balloon M. Hayashi and 0. Yokoyama 128 23. NCAR Boundary Profiler System R. B. McBeth and S. Semmer 136 Part 2 EXPERIMENT AND SUMMARY OF RESULTS 24. Details of the Experiment J. C. Kaimal, J. E. Gaynor, and H. W. Baynton 140 25. Summary of Results J. C. Kaimal, J. E. Gaynor, and H. W. Baynton 153 Figure 1.1. Instrumented 300-m tower at the BAO site. BAO SENSORS FOR WIND, TEMPERATURE, AND HUMIDITY PROFILING J. C. Kaimal NOAA/ERL/Wave Propagation Laboratory Boulder, Colorado, U.S.A. 1.1 INTRODUCTION The Boulder Atmospheric Observatory (BAO) is a research facility operated jointly by NOAA's Wave Propagation Laboratory and the National Center for Atmospheric Research (NCAR) . It is designed to provide high-quality measurements for use in boundary layer studies and instrument calibration. Located on gently rolling terrain 25 km east of Boulder, its main feature is a 300-m tower (Fig. 1.1) instrumented at eight levels to measure the mean and turbulent properties of atmospheric parameters such as wind speed, wind direction, temperature, and humidity. In-situ and remote sensors deployed around the tower provide additional information on the structure of the flow (e.g., pressure fluctuations at several stations, temperature structure from acoustic sounders, acoustic Doppler winds and wind convergence from a triangular configuration of optical crosswind sensors). A computer at the site processes information from these sensors for real-time printout of data summaries and archiving of raw data. Mean profiles of wind, temperature, and humidity for successive 20-min periods are available routinely. This real-time processing capability was an impor- tant factor in the choice of a site for the Boulder Low-Level Intercomparison Experiment (BLIE). A brief description of the tower sensors, the handling of data, and the parameters listed on the summary sheets can be found in the sections to follow. For more details of the site, instrumentation, and data processing see Kaimal (1978). 1.2 TOWER INSTRUMENTATION For BLIE, data from the eight instrumented levels on the tower (10, 22, 50, 100, 150, 200, 250, and 300 m) will be used as reference for comparing results obtained from the various sounding techniques. At each level, the sensors are distributed between two booms, one of which points approximately SSE (154°) and the other approximately NNW (334°) (Fig. 1.2). The three-axis sonic anemometer and the fast- and slow-response temperature sensors are mounted on the SSE boom (Fig. 1.3). In the absence of strong downslope winds from the Rocky Mountains, the winds blow generally from the southeast during the day, but from the west or northwest during the night. The orientation of the SSE boom is therefore optimum for daytime observations. A propeller-vane anemometer and a cooled-mirror dew point hy- grometer are mounted on the NNW boom (Fig. 1.4). 1.2.1 Three-Axis Sonic Anemometer The sonic anemometer used on the BAO tower uses a fixed orthogonal array; the two horizontal axes are oriented along and perpendicular to the boom, while the vertical axis is mounted at the end of the boom directed away from the tower (Fig. 1.3). The acceptable azimuth range for vertical velocity measurement is much larger in this array than in the non-orthogonal arrays described by Mitsuta (1974) and Kaimal et al. (1974). However, the horizontal wind measurements fare less well in this arrangement. Distortions in the flow within the acoustic path due to the presence of transducers cause the velocity readings to be underestimated when the wind direction is close to one of the anemometer axes. An approximate form for the velocity underestimation as a function of 8 (angle between the wind direction and the acoustic path) has been obtained from wind tunnel and atmospheric tests. The measured velocity component u approaches the true velocity component u t in the range 75° > 6 > 90° but drops off linearly with for angles less than 75° (Fig. 1.5). For a path length-to-diameter ratio of 25, appropriate to the BAO array, Figure 1.2. Details of the tower structure and booms. For a path length-to-diameter ratio of 25, appropriate to the BAO array, u = u (0.87 + 0.13 6/75) m t < 9 £ 75 '! 75 < 6 < 90 .1 (1.1) The correction is made on each data point sampled. While an arc tangent routine can be used for a first order approximation of 9 (with wind components measured along the two horizontal axes), it is more efficient to compute u t and v t using analytic expressions given below that very nearly approximate (1.1). Defining u r as the component more closely aligned with the wind and V)- as the component normal to it, u = C u t m v = D v t m (1.2a) (1.2b) where C = 0.926 + 0.31/(R + 1.38) D = 2.112 - C ; D > 1 =1.0 ; D < 1 (1.3a) (1.3b) R = v / u m m (1.3c) No such correction is needed for the vertical velocity measurements. A detailed discussion of these and other aspects of sonic anemometer performance can be found in Kaimal (1979). The sonic anemometer probes used on the tower include a mix of EG&G (Model 198-2) two-axis probes, Ball Brothers (Model 125-198) two-axis probes, and Ball Brothers (Model Figure 1.3. Three-axis sonic anemometer and slow-response quartz thermometer in aspirated shield on SSE boom. Figure 1.4. Propvane and dew point hygrom- eter on NNW boom. 125-197) single-axis probes. Driver and receiver circuits associated with the probe are located in waterproof boxes at each level on the tower; coaxial cables connect these cir- cuits to the computer interface circuits located in the van at the base of the tower. To insure proper synchronization in data-sampling the transmitters at all levels are fired simultaneously. The firing proceeds at 200 Hz, exactly 20 times the sampling rate (10 Hz) for all the fast-response channels. The interface circuit automatically accumulates the readings from 20 successive transmissions before transferring the data to computer memory. This block-averaging is provided to minimize aliasing in the spectral computations. 1.2.2 Propeller-Vane Anemometer A ruggedized version of R. M. Young Co . ' s propeller-vane anemometer, Propvane Model 8002, is used. The polystyrene propeller in this model has a distance constant of 2.4m and a working range from 1 to 54 m/s. Calibration of the wind speed output is ac- complished by driving the propeller shaft at a known rate of rotation (1800 rpm) and ad- justing the voltage level to the corresponding wind speed indication (15 m/s). The vane position is indicated by a precision conductive plastic potentiometer connected to the vane shaft. The potentiometer has a deadband of 18°. This deadband is pointed in the direction of the tower so that the active range of the potentiometer coin- cides with the azimuth range of best exposure for the NNW boom. Data for azimuth wind directions between 110° and 180° are often degraded by a combination of the tower inter- ference and the voltage jump produced by passage of the potentiometer brush across the deadband . 1 .2.3 Fluctuating Temperature Sensor A platinum wire thermometer (AIR Inc. Model DTIA) , mounted within the frame of the sonic anemometer's vertical axis probe, measures fluctuations in the temperature with a 80 -60 -40 •20 20 40 60 80 9 (degrees) Figure 1.5. Attenuation in measured sonic wind velocity component from flow distortions along the acoustic path caused by the transducers. The heavy line corresponding to d/a = 25 is the response curve for the BAO sonic anemometer. frequency response comparable with the path-averaged response of the sonic anemometer. The sensor consists of a length of 12-|J platinum wire (nominal resistance of 150 Q) wound around a helical bobbin. This wire is part of a simple bridge circuit the output of which is amplified by a low-noise, low-drift circuit to provide an output voltage of + 10 V corresponding to a temperature range of + 50°C. 1.2.4 Slow-Response Temperature Sensor Absolute measurements of air temperature are made with a Hewlett Packard quartz thermometer (Model 2850A) housed in an R. M. Young aspirated shield (Model 43404). The sensor is accurate to within 0.005°C, but has a long time constant, about 1 min. Calibra- tion checks performed over the years in the same precision temperature bath show no signi- ficant variation in the thermometers' calibration constants with time. 1.2.5 Dew Point Sensor A Cambridge Systems aspirated cooled mirror hygrometer (Model 110-SM) , calibrated as recommended by the manufacturer, measures the dew point at each level. The temperature of a mirror, made to track the dew point (or frost point) temperature, is measured. The cycling time for temperature control is approximately 1 s. 1.3 DATA ACQUISITION AND DISPLAY Acquisition of data from the sensors is controlled by a PDP 11/34 computer at the BAO site. The data acquisition system samples the fast-response sensors at the rate of 10 BOULDER ATMOSPHERIC OKSCRUA TORI DATA SUMMARr AVERAGING F'ERIOD= 20.00 MIN i'CM.l OWES osou u U OH AZ fOS 10 -3..?:' -3.44 -1 .69 4.41 -0. 18 4.73 43. 4.42 22 -3.37 -3.95 - 2 . 07 4. 76 -0. 10 5.19 41. 4.88 50 -3.44 -4,00 -2. 16 4.88 -0.18 5. 33 40. 4.98 100 -3.82 -4.07 - 1. , 98 5.22 ■,'.08 5 . 58 43. 5.03 ISO -3.96 -3,76 • 1 .64 5.20 . J 6 5. 44 46. 5,24 200 ■■3.°4 -3 . 33 -1.27 5 . 00 -0.05 5 . 1 5 50. 5.23 250 -4.11 -3.65 -1,48 5 , 30 -0.11 5.50 48. 5.24 300 -4.00 3.88 - 1 .73 5.29 . 1 1 5.5/ 46. 5.50 SEP 79 10 20 MSI 28 19 1 30 .'7 .'.-1 63 27 12 43 24 59 40 26 04 -0 20 25 48 49 25 04 -0 99 .'•1 54 -1 54 OPTICAL TRIANGLE RMS PRESSURE UALUES CMICR0BARS3 CM/SEC] A2 CDEG: 1.25 13. CONV ci/sec: 0.01118 L0G10(CN2) -13. 14339161 PRESSURE C MB 3 842.25 STN 1 STN 2 STN 3 6.461 7,832 6.597 SOLAR RADELY/MINl 1 .07 -25.77 -30.21 -63.04 -80.50 999.99 999. 99 999.99 999.99 10 1 .4315 0.9045 0.2380 0.3675 -0.41 58 -0,0337 -0.2360 -0.0999 -0.1314 0.1423 22 1.2687 0.7045 0.3757 0. 1983 - 0.3042 -0.0682 0. 1432 -0.0811 -0. 1500 0.1485 50 1 .0214 0. 525 7 0,6504 0.0 778 -0.2340 -0.0809 -0.0946 -0.0239 -0.2118 0. 1352 100 0.7430 0.6420 0,7717 0.0548 -0.0233 -0.0877 -0.0856 -0.0195 -0.2118 0. 1057 150 0.8567 0.7280 0,7761 0.0285 0.0889 -0.0716 -0.0497 -0.0184 -0. 1297 0.0727 300 0.9299 0.7570 0.6753 0.0211 -0.0354 ■-0.0B59 -0,0658 -0.0391 -0.1253 0.0454 250 1 . 1671 0.8311 0.7471 0.0284 0.0294 -0,0433 -0.0450 -0.0707 -0.2243 0.0325 300 0.8392 0.8532 0.5918 0.0264 - 0, 1268 0.0473 -0.0124 -0.0431 -0.0995 -0,0084 STN 4 STN 5 19.247 8.490 Figure 1.6. Sample listing of summary data for a 20-min averaging period. The time indicated at top right refers to the beginning of the averaging period. Column heads are explained in Table 1.1. Hz and the slow-response sensors at 1 Hz. Real-time computations of means, variances, covariances, and Obukhov lengths are made for consecutive 20-min periods (starting on the hour) and are listed on the line printer at the end of each averaging period. The summar- ies are recorded on magnetic tape along with a compressed form of the time series. To minimize tape storage requirements, only 10-s averages and 10-s grab samples (last sample in each 10-s averaging period) of the fast-response time series are stored. On the slow-response channels no attempt is made to save the 10-s grab samples. High- frequency information in the form of smoothed spectral estimates (approximately 10 esti- mates per decade) is stored for later use in extending spectra computed from the 10-s averaged data points. For this intercomparison experiment only the results from the summary sheets will be used. A sample listing for one period is shown in Fig. 1.6; Table 1.1 explains the symbols. The parameters relevant to this experiment (VWES, VSOU, W, VH , AZ , T, and T ) will be listed separately for comparison with data from systems operated by other participants. To minimize possible tower influence on the wind data, sonic anemometer readings will be used only for azimuth wind directions 64° - 154° (CW) . For wind directions 154° - 64° (CW) the Propvane readings will be used instead. Table 1.1. Explanation of terms used on BAO data summary sheet (Fig. 1.6) Z(M) Height (meters) AVERAGED PARAMETERS (SI units) VWES vsou u V w VH AZ PVS PVD T TD Horizontal wind component from west (sonic) Horizontal wind component from south (sonic) Horizontal wind component along the x axis, Horizontal wind component along the y axis, Vertical wind component (sonic) Horizontal wind speed (sonic) Horizontal wind direction (sonic) Horizontal wind speed (propvane) Horizontal true wind direction (propvane) Temperature (quartz thermometer) Dew point (dew point hygrometer) Obukhov length 2nd MOMENTS (SI units) UU, VV, WW, TT, UV, UW, UT , VT , UW, and WT (=u) Longitudinal wind component (sonic) ( =v ) Lateral wind component (sonic) (=w) Vertical wind component (sonic) (=6) Temperature (platinum wire) OPTICAL TRIANGLE (SI units) V Wind speed AZ Wind direction (CW) CONV Convergence LOG10(CN2) Structure parameter for refractive index x 10 12 RMS PRESSURE VALUES (microbars) (STN 1 . . . STN 5) PRESSURE (mb) SOLAR RAD ( Langleys/min) from 154° (sonic) from 64° (sonic) Referenced to 10 m mean wind direction From measurements along legs of an equilateral triangle, 450 m on each side, centered on tower From pressure fluctua- tion stations around the tower Mean surface pressure Mean solar radiation (direct and diffuse) 1.4 REFERENCES Kaimal, J. C. (1978): NOAA instrumentation at the Boulder Atmospheric Observatory. Prepr. Vol. 4th Symp . on Meteorol. Obs . and Instrum., 1978, Denver, Colo., American Meteorological Society, Boston, Mass., pp. 35-40. Kaimal, J. C. (1979): Sonic anemometer measurement of atmospheric turbulence. Proc. Dynamic Flow Conf., 1978, Baltimore, Md . , DISA, P.O. Box 121, Skovlunde, Denmark, pp. 551-565. Kaimal, J. C, J. T. Newman, A. Bisberg, and K. Cole (1974): An improved three-component sonic anemometer for investigation of atmospheric turbulence. In Flow: Its Measurement and Control in Science and Industry , Vol. 1, Proc. of a Symp., 10-14 May 1971, Pittsburgh, Pa., Rodger B. Dowdell (ed.), Instrument Society of America, Pittsburgh, pp. 349-359. Mitsuta, Y. (1974): Sonic anemometer-thermometer for atmospheric turbulence measurements. In Flow: Its Measurement and Control in Science and Industry , Vol. 1, Proc. of a Symp., 10-14 May 1971, Pittsburgh, Pa., Rodger B. Dowdell (ed.), Instrument Society of America, Pittsburgh, pp. 341-348. 2. SINGLE-HEAD SONIC ANEMOMETER-THERMOMETER Tatsuo Hanafusa Meteorological Research Institute Japan Meteorological Agency Tsukuba, Japan Yasuhiro Kobori Kaijo Electric Co. Tokyo, Japan Yasushi Mitsuta Disaster Prevention Research Institute Kyoto University Kyoto, Japan 2. 1 INTRODUCTION A sonic anemometer-thermometer detects by sonic means the wind velocity component and the air density or temperature along the sound path. The first successful sonic anemometer-thermometer for use in the atmospheric turbulence studies was developed by Suomi (1957). In his instrument, velocity and temperature fluctuations were obtained by the difference and sum of transit times for two series of sound pulses traveling in oppo- site directions along the same sound path. Amplitude variations in the received pulse signal of Suomi ' s instrument caused errors in transit time detection. To avoid such errors, Kaimal and Businger (1963) of the University of Washington developed a continuous-wave sonic anemometer-thermometer which measured phase shifts between the two signals received at either end of the sound path. Their instrument gave satisfactory performance in field observations. Improvement of the pulse-type instrument continued at Kyoto University, and a practical instrument was completed by Mitsuta (1966). The instruments, developed independ- ently by the University of Washington and Kyoto University, when compared in the United States in 1965 showed excellent agreement in the measured wind velocity component (Businger et al. , 1969) . After discussions of the merits of the instruments, the two groups decided to develop a new three-dimensional sonic anemometer-thermometer of pulse type as a joint effort. The pulse-type instrument appeared more promising for absolute measurement of pulse transit times and for attaining wide observation range and stable zero point, if only a reliable means of transit-time detection could be developed. Subsequent development of a new technique to determine the time of reception of the signal pulse (by making the trigger pulse from the third wave of received signal pulse instead of the wave envelope) enabled the Japanese group to produce a new solid state three-dimensional sonic anemometer- thermometer (Mitsuta et al., 1967). After several improvements a more refined version with three 20-cm sound paths and using 100-kHz acoustic pulses was completed (Mitsuta, 1974). It was widely and successfully applied in various field experiments as reported by Japan-U.S. Joint Study Group (1971) and has been produced commercially in Japan. Through almost a decade of practical application of this instrument, we have identified major sources of unreliability. One important source was zero-point drift caused by small variations in the geometrical length of sound paths. Small differences in length between two opposite sound paths cause large zero offsets in the two-head type anemometer. The other source of unreliability arose from the temperature and cross wind approximations in the derived wind and temperature. The new single-head sonic anemometer- thermometer described by the present authors achieves more reliable measurement by removing Figure 2.1. The new single-head anemo- meter-thermometer: a) wind antenna specifications; b) main chassis, junc- tion box, and wind antenna (from left to right) . the difficulties mentioned above. The basic points of improvement from the 1971 model are adoption of a more precise formula for processing, a single-head two-way sound path, and a more effective noise-gate circuit. 2.2 BASIC PRINCIPLES The transit time of a sound signal traveling from one end of the sound path to the other separated by distance d can be written as follows (Schotland, 1955): (C t = v 2 ) 1/2 + v n — d (2.1) where V is total wind velocity, V^ and V_ are wind velocity components in the directions parallel and normal to the sound path, and C is the velocity of i^ound in still air. The sign, +, before Vj should be chosen according to the direction of sound transmission. If two transit times t 1 and t 2 in opposite directions on the same sound path are detected, V^ can be obtained independent from V and V n as follows: V =4 (i. d 2 V (2.2) C can be obtained by assuming C >> V as follows: n C =| (f- + f-) • (2-3) 1 2 The velocity of sound in still air, C, can be expressed as follows (Barret and Suomi , 1949): C = 20.067T 1/2 , (2.4) sv where T sv is sound virtual temperature of the air and is equal to air temperature if the atmosphere is dry. In moist air T gv can be correlated with air temperature, T, by water vapor pressure, e, as follows: T = T(l + 0.3192e/p) (2.5) sv r where p is atmospheric pressure. Then T gv can be obtained from two transit times t 1 and t 2 by the following equation: T = ^ j ( L ~ + W . (2.6) SV (2 x 20.067r t l t 2 In the present instrument, the pulse transit times are processed electronically following (2.2) and (2.6) to obtain the wind velocity and virtual temperature outputs. The wind velocity component obtained by equation (2.2) is the line-averaged wind velocity component over the sound path of length d in the direction parallel to it; air temperature obtained by (2.6) is also sound virtual temperature averaged over the sound path, d. Those are apparently smoother than the point-detected entities. The attenuation characteristics of this line averaging were studied by Mitsuta (1966) and in more detail by Silverman (1968) . 2.3 DESCRIPTION OF THE INSTRUMENT The new three-dimensional sonic anemometer-thermometer is shown in Fig. 2.1, and a block diagram of its functions is shown in Fig. 2.2. The sensor (wind antenna) has two horizontal sound paths (A,B) crossing at 120° and a vertical sound path (W) ; at both ends of each path are two-way sound transducers which perform alternately as transmitter and receiver. The switching circuit, transmitter, and pre-amplif iers are built in the junction box located near the sensor. The main chassis consists of a main oscillator, a main controller, transit time detection circuits, wind velocity converters, a temperature converter and a power supply. All components are solid state, many of them being inte- grated circuits. The main oscillator is a crystal oscillator of 5.5556 MHz. It supplies the clock signals for wind calculation (5.5556 MHz), and for temperature calculation (2.7778 MHz), and the timing signal to the main controller which generates timing signals (110 Hz, six times the observation cycle) to switch transducers and to trigger noise gates and converting circuits. The lead-zirconate transducers at both ends of the three sound paths (A,B,W) are triggered alternatively in the sequence of A + , A - , B + , B _ , W and W~ at equal TRANSIT TIME SIGNAL GEN. INVERSE CALCULATION WIND COUNTER MAIN CONTROLLER I MAIN OSCILLATOR CALIBRATION TEMPERATURE COUNTER 1 SQUARE CIRCUIT TEMPERATURE COUNTER 2 D/A CONVERTER Li, J OUT OUT OUT T A Y B ' DIGITAL ANALOG WIND OUT WIND OUT 300V +15V-15V POWER SUPPLY w D/A CONVERTER I OUT J ANALOG DIGITAL TEMP OUT TEMP OUT Figure 2.2. B thermometer . AC 100V lock diagram of the new single-head anemometer- intervals, clocked by the timing signal. The transducers, when triggered, send out burst signals at resonant frequency of 100 kHz. When the transducer at one end of the sound path is triggered, the transducer at the opposite end of the path switches to its receiving mode . The received signals sent to the receiver amplifier are first separated from the noise by noise gates, then shaped and triggered at the zero crossing point of the third wave to generate the receipt time signal pulse. The transmitting pulse delayed to adjust for the zero point and the receipt time signal pulse are supplied to the inverse calcula- tion circuit which produces the signal pulse whose width is proportional to the inverse of the pulse transit time of each sounding. The counter for the wind velocity component is an up-down counter clocked by 5.5556 MHz to make the difference of width of the signal pulses corresponding to the 10 Table 2.1 . Specifications of the new sonic anemometer-thermometer Anemometer Thermometer Measuring mode Time-sharing multiplex transmission/reception switchover type ultrasonic pulse emission, ~ 20 Hz per channel. Measuring range "*• +30 m/s Central temperature: -10° "- A0°C Temperature deviation: "" +5°C Accuracy 1% 1% Minimum resolution 0.5 cm/s 0.025°C Frequency resolution 10 Hz 10 Hz OUT 1: to +1 V/lOm/s 8 V max OUT 1: to +1 V/+50°C OUT 2: to +1 V/full scale OUT 2: to +1 V/+5°C Analog output Full scale *U:+5, +10, +25, +50 m/s **W:+1, +2, +5, +10 m/s Digital output'-'" 1 '"" 15 bit binary code 12 bit binary code Operating temperature Main unit: -10° to A0°C, Probe and junction box: -20° to 50°C Power supply AC 100/115/220 V +10% 50/60Hz *U: Horizontal component. **W: Vertical component. '-'-''Not used in the present intercomparison . successive two soundings in the opposite directions on the same sound path, such as A and A - , B and B~ , and W and W . One bit in this counter corresponds to 0.005 m/s. The digital signal corresponding to the difference is converted into a voltage analog signal in a D-A converter and then distributed to each wind component output circuit. The first counter in the temperature circuit produces a voltage analog signal proportional to the sum of the two successive soundings in opposite directions on the vertical sound path (W + W~) . This voltage is squared, producing a temperature signal pulse whose width is proportional to sound virtual temperature. The width of the tempera- ture signal pulse is converted into a temperature digital signal by the second counter clocked by a 2.778 MHz signal from the main oscillator. One bit in the temperature cor- responds to 0.025°C. The digital signal is converted to an analog temperature output through a D-A converter. Digital techniques were adopted in the pulse-width detecting circuit to attain high accuracy and stability of measurement. Hybrid systems of digital and analog circuits have been replaced by pure digital processing systems. 2.4 PERFORMANCE All of the circuit parameters can be tested and adjusted electronically within their limits of tolerance, except the aeroacoustic characteristics of the wind antenna. The specifications of the new sonic anemometer-thermometer are as shown in Table 2.1. The aeroacoustic characteristics of the wind antenna were tested in the wind tunnel of MRI . The length of the sound path is 20 cm, and the diameter of the transducer is about 1.5 cm. Therefore the wake to the lee of the transducer produces some errors in measurement even though streamline shape of the transducer is improved. The horizontal wind speed component detected by each leg of horizontal sound path for various wind direc- tion to the wind antenna is as shown in Fig. 2.3. The crossing angle of the leg is 120°, 11 U(m/s) • ' u A, • • -5.0 -4.0 .*' -.3.0 " • . ' • • (c fV I' • •*.// v.. A. //4000 m with better capabilities (pay- load, ceiling, flight duration, aerodynamic self-stability) than the SAM-B. It allows for simultaneous pressure, temperature, and humidity measurements and gas samplings. 4.2 SYSTEM DESCRIPTION 4.2. 1 SAM-B SAM-B has a circular wing of polystyrene covered with kraft paper; the body (moulded fiber) contains the instruments. Its "critical aerodynamics" permit a parachute- like descent, providing an easy and safe approach and landing, as well as good aerodynamic stability at low speed (10 m/s). The simplicity of the design provides low cost and easy maintenance. Figure 4.1 shows SAM-B taking off. Limitations of this aircraft are as follows : (1) It has a small visual range between pilot and aircraft (800 m) because of its size and geometry. (2) Maximum wind it can be used in is 10 m/s. (3) Operational ceiling is about 2000 m. To overcome these limitations, a SAM-D (Fig. 4.2) with a delta wing that has a front stabilizer is currently being developed, allowing a wider speed range, a vertical speed of about 7 m/s, an operational ceiling of 5000 m, and the ability to be used in winds up to 15 m/s. Characteristics of SAM-B and SAM-D are given in Table 4.1. 4.2.2 SAM-C SAM-C is a self-stabilizing motorized glider with modular design, made of poly- styrene and fiberglass. A special feature of the aircraft (Fig. 4.3) is its inverted "V" Figure 4.1. SAM-B taking off. Figure 4.2. SAM-D, a new vehicle being developed . tail, giving low drag (two wings instead of three) and especially easy control during high- altitude flights. Tail retraction at landing is achieved by use of a semi-rigid hinge. The weight of SAM-C and the need for operation on rough ground dictate the use of a cata- pult and recovery nets. Because of a fineness ratio of 10, SAM-C can be operated in strong winds (15 m/s). Technical specifications are given in Table 4.1. 19 Table 4.1. Technical characteristics of SAM aircraft SAM-B (Circular wing) SAM-C (Classic airplane: reversed dihedral) SAM-D (Stabilizer in front of delta wing) Payload 0.6 kg 1.8 kg 0.8 kg Total weight 3.5 kg 7.5 kg 5.2 kg Minimum speed 5 m/s 10 m/s 9 m/s Cruise speed 15 m/s 25 m/s 30 m/s Initial climb rate 4 m/s 5 m/s 7 m/s Operational ceiling 2000 m 4000 m 5000 m Endurance at sea level 15 min 1 h 20 min Wing span 1.20 m 3.15 m 1 .40 m Wing area 0.95 m 2 1 m 2 1.75 m 2 Air foil NACA 4415 NACA 4415 Engine (diesel) 10 cm 3 1.5 HP at 12000 rpm Propeller 28 cm x 18 cm Fuel Methanol + nitromethane + oil Figure 4.3. SAM-C taking off. Its weight (8 kg) dictates the use of a catapult. 20 Sensors Converters Modulation (Standard IRIG) p F / / F Channel E 1500-3000 Hz 70 kHz Antenna T R / F Channel C Mixing Amplifier FM Transmitter 403 MHz /I V750-3000 Hz 40 kHz Vj U R F Channel A 750-3000 Hz 22 kHz Figure 4.4. Block diagram of telemetry. Antenna V 7 A/D Converters Receiver t Filters , 1 . 1 ' ~ i F / /BCD F /BCD F / /BCD 1 ' 1 1 r~ ~~ Interfacing ♦ Plotting Board Processing Unit Printer * * Recorder Figure 4.5. Block diagram of data acquisition. 4.2.3 Remote Control Digital remote control uses a 250-mW transmitter with a center frequency at 436 MHz and a bandwidth of 50 kHz. This device controls the engine, the airbrakes, the aile- rons, and the gas-sampling system. 4.2.4 Telemetry Figures 4.4 and 4.5 are block diagrams of the telemetry and data acquisition. Pressure, temperature, and humidity are measured with a vibrating wire sensor, thermistor, and hygristor (Fig. 4.6). Their characteristics are given in Table 4.2. Temperature and 21 Time (s) AQ A (J ® AQ P Figure 4.6. Aircraft instrumentation. The picture shows the pressure sensor, the micropump, two filters for gas sam- pling, and the telemetry equipment. Figure 4.7. Data processing sequence: AQ) data acquisition, 1) median compu- tation, 2) transfer function (fre- quency/parameter), 3) recording of median value, 4) plotting on 9872 A. Table 4.2. Characteristics of PTU sensors Sensors Measurement Range Resolution Precision Pressure (vibrating wire) Temperature (thermistor ML 419] Humidity (hygristor - ML 476) 150 mb between 1050 and 500 mb •15°C to + 40°C 20% to 100% 7 Hz/mb 20 Hz/°C 20 Hz/% +0.1 mb +0.2°C +5% humidity sensors are protected against solar radiation, and the shielding device offers appropriate ventilation during the flights. The transmitter has a carrier frequency of 403 MHz, an HF power of 60 mW, a bandwidth of 100 kHz, and IRIG subcarriers of 22 kHz (humidity), 40 kHz (temperature), and 70 kHz (pressure). It is thus possible to get real-time profiles of temperature and humid- ity versus pressure or approximate altitude, assuming that the pressure gradient is con- stant within the flight range. 4.2.5 Data-Processing The data-processing system consists of a Hewlett Packard HP 9825 minicomputer (24 K memories), a plotter, and a 9866B thermal line printer (Fig. 4.5). The system allows for plotting one value every 3 s, corresponding to three measurements (acquisition rate = 1 Hz). The computer eliminates extreme values, retains the center value, converts into meteorological data, plots temperature versus pressure, and stores the data on a cassette unit. (Figure 4.7 is a time diagram of the computer operations.) 22 940 1 \ 950 960 '•• 970 V Pressure (mb) CD CD CD OO o o - V* 1000 - 1010 - V - 1020 1 nr>n - 1 - -10 +20 Temperature (°C) • 30 Figure 4.8. Example of temperature-versus- pressure plot on new emagram '900' designed for the boundary layer. (Plot shown is redrawn from original real-time plot.) Taken at Porcheville 6 July 1979 at 0600 by SAM C06 . 4.2.5.1 Real-time proces sing Real-time plotting of temperature versus pressure is possible on any emagram. A specially designed one (Fig. 4.8) for low atmosphere is characterized by a linear pressure scale between 1050 and 900 mb and isotherms perpendicular to isobars. 4.2.5.2 Delayed data processing (20 min) Stored data can be further processed. The raw data extraction program prints time, pressure, temperature, humidity, height (through Laplace equation), and altitude. The correction program eliminates the wrong values from the first shorter set. The final processing program prints time, pressure, temperature, humidity, height, altitude, dew point, adiabatic wet-bulb temperature. Final data are also recorded on cassettes. 4.2.5.3 Plottings Processed data can be plotted on emagrams as temperature, dew-point temperature, and adiabatic wet-bulb temperature versus pressure (Figs. 4.9 and 4.10). Certain flight parameters, such as pressure or vertical speed versus time or height (Fig. 4.11), can also be plotted. 23 800 850 900 Q. 950 1000 1050 + 15° Temperature (°C) 25 Figure 4.9. Example of T, T, and 'W plots during ascent. (Plot is redrawn from real-time plot, which shows points for both ascent and descent.) Data were taken at Chateauroux 23 July 1979 at 0500 by SAM Bll. 4.3 SUMMARY OF PREVIOUS USES 4.3.1 Mount Etna A remotely piloted SAM-C was successfully used in June 1978 on Mount Etna, a volcano in Sicily, in cooperation with H. Tazieff's team of volcanologists . The experiment had two objectives: (1) To study the thermal environment of the volcano to assess the modification of the vertical profile of temperature by the mountain (and the volcano). (2) To sample the gases in the plume of the volcano to determine chemical properties and prove the feasibility of the SAM method for physical and chemical research of volcano plumes. 4.3.2 Chateauroux SAM-B was employed at Chateauroux during July 1978 to monitor the inversion layer early in the morning and give accurate knowledge of the time of thermal plume forma- tion, so that take-off schedules for the Glider World Championship held there could be 24 800 600 400- 200- 790 800 810 - -820 830 10 15 Temperature (°C) 2 b 30 Figure 4.10. Example of T, Tj, and RH plots during ascent as a function of height. (Plot shown is redrawn from real-time plot, which shows points for both ascent and descent with different measurements indicated in different colors.) Taken at Boulder 28 August 1979 at 0930. 750 != 800 850 1000 Figure 4.11. Pressure versus time plot recorded during flight. fixed. Thirty-eight soundings were performed during 14 days, with time intervals varying from 1 h (around 0500 UT) to 20 min (around 1000 UT) . 4.3.3 Plant Pollution Studies During July 1979, SAM-C was used with a new sampling device consisting of 1.5-£ glass containers with an initial internal pressure of 1 mb . These containers were opened by a remote control system and were filled in approximately 5 s, then closed again. Anal- yses were conducted on the site immediately after each flight with an ultraviolet fluores- cence method (S0 2 ). 25 5. COMPARISON OF AIRCRAFT AND BAO TOWER MEASUREMENTS D. H. Lenschow and B. B. Stankov National Center for Atmospheric Research Boulder, Colorado, U.S.A. 5. 1 INTRODUCTION From 17 to 28 April 1978, a site evaluation study was conducted in the vicinity of the Boulder Atmospheric Observatory (BAO) tower to help determine the degree to which the tower measurements are representative of the surrounding area. This was the initial field research program involving the BAO tower. An NCAR Queen Air aircraft was used to obtain measurements along horizontal flight paths centered at the tower for comparison with concurrent tower measurements. Intercomparisons of tower and aircraft measurements are carried out periodically by NCAR primarily to calibrate and check aircraft measurements of temperature, pressure, and humidity. This is particularly important for experiments involving several instrument platforms probing the same phenomena. Biter and Wade (1975) describe the techniques and equipment used for such an intercomparison during the National Hail Research Experiment (NHRE) and present some examples of the results. Their accuracy tolerances were +0.5 K for air and dewpoint temperature, and +1 mb for static pressure. Measurements of turbu- lence quantities such as variances and fluxes, however, are not routinely compared. The region surrounding the tower is gently rolling terrain used for grazing, dryland and irrigated farming, and suburban living. As shown in Fig. 5.1, the general terrain slope is toward the northeast, although the local slope near the tower is toward the north. As pointed out by Lenschow et al. (1979), even very slight terrain variations can cause large variations in the overlying boundary layer, particularly at night and during the initial development of the daytime convective mixed layer. In addition to local terrain variations, the foothills of the Rocky Mountains rise to more than 600 m above the BAO site within 25 km west of the tower. About 25 km farther west, the Front Range towers about 2500 m above the BAO site. Therefore, we expect that, under certain conditions depending upon the height above the ground, the tower measurements will not represent the surrounding area. 5.2 EXPERIMENTAL DETAILS The BAO tower was instrumented for measurements of wind velocity components, temperature, and humidity at heights of 10, 20, 50, 100, 150, 200, 250, and 300 m above the ground. Details of the tower, its instrumentations, and the data acquisition and recording are discussed by Kaimal (1978) . Twenty-minute-block averages of winds and temperature, recorded at a sampling rate of 10 Hz, and standard deviations from the 20-min means were calculated and compared with the aircraft measurements. The NCAR Queen Air, shown in Fig. 5.2, is a light twin-engine aircraft that can be equipped for a variety of research programs (Burris et al., 1973). The aircraft meas- urements in the site evaluation study included the three wind components, air and surface temperature, and humidity. Air motion measurements were calculated from the airplane velocity and attitude angles, measured with an inertial navigation system (INS), and the air velocity with respect to the airplane was obtained from sensors at the tip of the nose boom. A Rosemount platinum-resistance-wire thermometer, with a wire diameter of 25 |Jm, is also mounted near the tip of the nose boom. The measured total air temperature is then corrected for heating caused by the rapid air flow. Further details on the air motion sensing system are presented by Lenschow et al. (1978), and on airplane temperature and velocity measurements by Lenschow (1972). 26 40 2 40 I 40.0 399 105.0 104.9 104.8 104.7 Figure 5.1. The region surrounding the Boulder Atmospheric Observatory tower and the research aircraft flight path ( ). The aircraft data were recorded at 20 Hz and divided into segments of 4096 sam- ples. At an average aircraft speed of 70 m/s, this is equivalent to a length of 14.3 km. For comparison with the tower data, at a wind speed of 10 m/s, the equivalent averaging distance for a 20-min time series is 12 km. Before standard deviations were calculated, the mean and a least-squares linear trend were removed from each segment. Because of flight restrictions over the suburban tower surroundings, the minimum aircraft flight altitude is 150 m above the ground. Therefore, intercomparison measure- ments were made only at the 150- and 300-m levels of the tower. The aircraft flight path is shown in Fig. 5.1. The aircraft was flown at a constant height above the ground, measured by a radar altimeter. Altogether, 13 flights totaling 28 h were conducted. Here we present a comparison of tower and aircraft measurements during Flight 6, from 1551 to 1751 MST on 21 April 1978. A weak cold front had passed through the area several hours earlier and the wind was quite steady from the northwest. The sky was partly covered with cumulus clouds, with no precipitation. Solar heating was sufficient to generate a convective mixed layer well above the airplane flight path. The flight path used on this day was a 30-km-long I-shaped pattern, centered at the tower, as shown in Fig. 5.1. This pattern, which took less than 15 min to complete, was flown continuously for eight cycles, all at 150 m above the ground, except for the last cycle, which was at 300 m. On most of the other days, an X-shaped flight path was 27 Figure 5.2. NCAR Queen Air research aircraft. chosen to allow comparisons of aircraft and tower measurements at locations approximately uniformly surrounding the tower. However, it took 30 mm to complete this pattern and, furthermore, on two of the four traverses by the tower, the airplane turned at the tower. Since turns can have an adverse effect on the accuracy of spatially averaged airplane wind measurements, reliable values centered at the tower were obtained for only half of the flight legs past the tower. 5.3 TOWER-AIRCRAFT INTERCOMPARISON Tables 5.1-5.3 summarize the comparisons between the airplane and tower measure- ments. Since the length of an entire flight leg is only about 30 km, the north (Table 5.2) and south (Table 5.3) segments each overlap the center segment by about 6 km. The mean values of tower and airplane measurements of horizontal wind components and temperature agree well with each other. The average difference of the velocity compo- nents is 0.6 m/s, with no significant difference between the segments by the tower and the other segments. Similarly, the average temperature difference is only about 0.4 K. Thus, horizontal variations were not large enough during this period to cause significant errors in the mean. The standard deviations of the aircraft horizontal wind components are consist- ently somewhat larger and more variable than the tower measurements, particularly for the segments away from the tower. Although we do not know the reason for this, there are two possibilities : (1) The wind may be affected somewhat by the terrain. On the north leg, we observed that u consistently reached a maximum in the middle of the segment, then decreased at the end of the leg. It is difficult to determine, however, even the scale of terrain variation that might cause such a perturbation. (2) The airplane measurements may include contributions from longitudinal rolls (LeMone, 1973) since the airplane flight path is approximately perpendicular to the wind. If the rolls are aligned approximately with the wind, the tower may not pass through a complete cycle during the 20-min measurement period. Tower and airplane measurements of the standard deviations of vertical velocity and temperature show good agreement with each other. This may be the result of relatively less contribution to the standard deviations by terrain-induced variations and longitudi- nal rolls than by the horizontal velocity components. 28 >> 4-1 ■H o o r-t 1j > IH O VI *J S-l Cj Oj 1) 5 E o 0) +j M 3 ^J >H H c 1-1 (/) (1/ 4h QJ S o O a CM CM en ^0 CM r^ en en on ^0 cn o> cn LO en - 1 ~ ' H o o - 1 " ^0 CM cn vo o \D CM ■- ■- 1 - 1 rH *" "- 1 ■- 1 O m cn 00 rl o o> r-- 1 LO 1 LO 1 i LO 1 LO 1 o en o r^ 00 in rt LO o r^ o r-> - CM *"* o 00 00 O en cn CO en CM CM o io o LO o LO o LO o LO O O en 11 O CM o CM a u OJ > < rH r^ o a; > S-l OJ i — i J^ K 4J 3 c d o N OJ ■H -C S-l 29 a o G u 0) U T3 d w d ,-— "V 0) < E « 4J OJ ft w •H !-. -j u XI w « - •H 3 — tn 4-1 ID i £ •V O d u s CvJ u-) 01 i— 1 X> &j £ •r 0, •H o 4-1 ^J •H a Sj a, 4H . O w 3 — C 3 w en *•— ' — 0) !-l S-. 0) 01 * -J c 4-) 11 u TJ a W - *-> a /-^\ V < E Ml *J 0) ft CO ■1 1j *J U ■d c/3 oo x: ■H 3 —J w M— t U Cb d a j-> ifl — i s- ft O S-l LH ■H CO 0) S-i ^ 3 ~-> M C !-. 3 11 W ft ■H £ S-l 01 Cfl 4-J ft E -r c CJ u s 11 r— I rC M H tj E T3 01 ■H O •U J-J ■^ H c ^ OT 01 4-1 Oj S c_> o O. o I d o o o i o u-i I o I c o o a CO O o o 31 5.4 SUMMARY The excellent agreement found in this initial comparison of mean values and standard deviations of velocity and temperature measured concurrently by the BAO tower and an aircraft is very encouraging. We can make useful comparisons of many other items of interest, such as covariances, time changes at the tower versus horizontal gradients from the aircraft, and spectral and cospectral quantities. These comparisons could help to determine the degree to which the tower measurements are affected by terrain inhomogenei- ties . Varying degrees of horizontal inhomogeneity in the boundary layer might be observed with different synoptic situations and at different times of day. Comparison of airplane and tower spectra may also be useful in determining the validity of Taylor's hypothesis, i.e., that time and space averages are interchangeable. 5.5 REFERENCES Biter, C. J., and C. G. Wade (1975): Field calibration and intercomparison of aircraft meteorological measurements. Preprint Vol. 1, NHRE Symposium/ Workshop on Hail, September 1975, VIII . A . 63 . 1-63 . 21 . (Available from NCAR , P.O. Box 3000, Boulder, CO 80307.) Burris, R. H., J. C. Covington, and M. N. Zrubek (1973): Beechcraft Queen Air aircraft. Atmos. Technol. 1:25-27. Kaimal, J. C. (1978): N0AA instrumentation at the Boulder Atmospheric Observatory. Prepr. 4th Symp . on Meteorol. Obs. and Instrum. , 10-14 April 1978, Denver, Colo., American Meteorological Society, Boston, Mass., pp. 35-40. LeMone , M. A. (1973): The structure and dynamics of horizontal roll vortices in the planetary boundary layer. J. Atmos. Sci. 30:1077-1091. Lenschow, D. H. (1972): The measurement of air velocity and temperature using the NCAR Buffalo aircraft measuring system. NCAR-TN/EDD-74. National Center for Atmospheric Research, Boulder, Colo., 39 pp. Lenschow, D. Fj. , C. A. Cullian, R. B. Friesen, and E. N. Brown (1978): The status of air motion measurements on NCAR aircraft. Prepr. 4th Symp. on Meteorol. Obs. and In- strum., 10-14 April 1978, Denver, Colo., American Meteorological Society, Boston, Mass. , pp. 433-438. Lenschow, D. H. , B. B. Stankov, and L. Mahrt (1979): The rapid morning boundary layer transition. J. Atmos. Sci. 36:2108-2124. 32 TETHERED AERODYNAMIC ALLY LIFTING ANEMOMETER (TALA)" Charles F. Woodhouse Approach Fish, Inc. Clifton Forge, Virginia, U.S.A. 6. 1 INTRODUCTION Variable altitude anemometry is a new method which is beginning to find applica- tion in low-level wind measurement. The principle is simple. Since wind-tunnel veloc- ities are used to calibrate the lift and drag of an airfoil, if the characteristics of a free-flying airfoil are known then the velocity of the wind can be determined. The air- foil sensor or kite described here is stable in naturally turbulent winds to 50 m/s (Fig. 6.1). Figure 6.1. Patented sled airfoil is stable to 40 m/s. The tether line force is practically that of a theo- retical flat plate: 1.967 pv 6.2 MEASUREMENT OF WIND SPEED The tension on the tether line produced by lift and drag forces acting on the kite provides a measure of wind speed at kite level. The force on a flat plate is P v (6.1) where q - force, p - density, and v = speed; q is quite close to the force on the sensor q, as determined by wind tunnel calibration: P v 1.967 (6.2) ■U.S. Pat. numbers 4,058,010 and 4,152,933. $3 > ,J3 c < < Figure 6.2. Original calibration of air- foil at NASA-Langley with correlation at NBS of electronic equipment built to Langley formula. • = NASA-Langley data points (calibration 5 April 1976; corre- lation coefficient .999). ♦ = National Bureau of Standards data points (cali- bration 17 March 1978; correlation co- efficient .997). 5 10 15 20 25 30 Wind tunnel true (m/s) The force q^ is measured directly by a strain gauge attached to the end of the tether line. Correlation coefficients as determined in the large NASA-Langley and the NBS wind tunnels are 0.999 and 0.997 respectively (Fig. 6.2). Car-tow calibration in still dawn air (Fig. 6.3) at 175 in altitude shows a wind drift between east and west runs. Car speed was calibrated by radar. Calibration in natural winds against cup or propeller anemometers is difficult because the empirical and varying turbulence correction equations of these instruments (McMichael and Klebanoff, 1975) are in the 1% to 2% range (Baker et al . , 1979). Pre- vious calibration against the BAO sonic anemometers indicates accuracies within 0.65%. We will test this agreement more accurately during the BLIE intercomparison. Catenary drag and the relationship between catenary cable curvature and height have been extensively studied. The present cable used is only 250 microns in diameter with a weight of 115 g/km. Computer analysis of the dynamics of this line (Shieh and Frost, 1979) has shown that for a cable of that type the maximum effect of catenary drag on the measured tension at the ground is five orders of magnitude smaller than the tension produced by the airfoil itself. 6.3 MEASUREMENT OF ALTITUDE The kite altitude is a function of line length, corrected for catenary sag and the observed vertical angle to the kite. The following empirical relationship is used: Z = 0.9 sin(J)(0.3n - 2.2 x 10 5 n 2 ) (6.3) where Z is the kite altitude, is the observed sensor vertical angle and n is the reel count (Fig. 6.4). The 0.9% correction factor in (6.3) compensates for the effect of catenary curvature (Shieh and Frost, 1979). This correction is small because of the low weight and low drag of the line. The equation used for estimating decrease in reel diam- eter as line is payed out is found to be accurate to 1%. Vertical angle (J) can be determined from a hand-held clinometer. In the elec- tronic model, altitude is computed automatically from measurements of the string angle at 34 20 15 - 8 1 ° o o 5 - Altitude 175m «/ Altitude 310m • / • East /% • East / X x West x West X •/ x X / • > X X / X / • X V x /• X i i 1 1 1 1 10 15 10 15 20 Car Velocity (ms- 1 ) Figure 6.3. Car tow calibration at 175 m shows slight wind drift between east and west runs. the base and of the line payout. A catenary correction control allows compensation for differences between visual angle and string angle, for different amounts of line length. The kite flies at a constant angle of attack to its apparent wind. Wind-tunnel data show a tether repose angle of 53° at 5 m/s rising to 57° at 20 m/s. Car-tow data (Fig. 6.5) illustrate the constant flight altitude at speeds of 5 m/s to 20 m/s. 6.4 MEASUREMENT OF WIND DIRECTION A simple potentiometer device attached to the end of the tether line measures azimuth wind direction. The potentiometer readings are referenced to true north. A pronounced lateral bend in the line will introduce errors in the wind direction reading, but this can be corrected by entering into the data acquisition system an initial visual reading of the direction of the upper section of the tether line. In order to make this measurement it will be necessary to move to either the left or right of the tether point with a precision magnetic compass. 6.5 CHARACTERISTICS OF THE KITE The airfoil weighs 16 g and has a surface area of 1500 cm . It acts, when under tension in flight, much like a large electrostatic loudspeaker cone. Because of the low weight and large surface, its compliance is high. The tether, or force transmission line, is of inelastic Kevlar (DuPont) with a modulus of elasticity in the range of steel. Thus, the frequency response of the sensor is high. Field measurements from the sensor flying at 100- to 200-m altitude show a response of at least 10 Hz in a 10 m/s wind field. The free-flying sensor orients directly and at constant angle of attack to its apparent wind. Thus its altitude of flight at specific line lengths is a direct function of the vertical component of the apparent wind vector. Similarly, horizontal position changes are a function of the lateral component. A measurement of the lateral and verti- cal turbulence may be obtained to altitudes of 300 m. 35 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 V = 10 m/s L = 360 m = 30.5 Approximately 0.1% Figure 6.4. Analysis (Shieh and Frost, 1979) showing 0.1% effect of varying wind field on altitude where v = wind speed at sensor level; L = line length; (J) = sensor elevation angle; Z = height; X = distance; p = power law exponent. The kite will fly stably at wind speeds as low as 3 m/s. To lift the kite under such conditions, an ancillary balloon inflated with helium to a specific circumference may be attached with surgical silk, or mercerized cotton thread to the sensor. The surgical silk breaks away at wind speeds of 4 m/s. At wind speeds of less than 3 m/s the helium-filled balloon lifts the sensor through calm inversions to upper altitude winds. The angle of the restraining tether is a function of lift/drag, and the altitude is a function of line length out and tether repose angle. As the sensor is lifted through the inversion into winds greater than 4 m/s the lifting balloon breaks away from the now flying sensor. 6.6 EQUIPMENT OUTPUT CHARACTERISTICS Wind speed: The line tension strain gauge outputs are internally calculated to give 400 mV/m/s; thus, 10 V = 25 m/s. Direction: Tether line direction is calculated from a reference centerline voltage of 5.0 V at + 36° per volt. 36 300 250 2 00 - 150 - 100 5d Reel Count 2340 J L 15 20 Velocity (m/s) Figure 6.5. Altitude of sensor at increasing line length and speed, illustrating relatively constant altitude of flight at velocities above 5 m/s. Altitude: Altitude from to 300 m is calculated from line length and catenary angle at 33 mV/m; thus, 10 V = 300 m. Digital: Analog oscillation-free outputs are sampled at 0.8 Hz by a portable digital recorder. Mean standard deviation and minimum and maximum values are calcu- lated for each parameter each minute and stored on a cassette tape for later data reduction. The data output is RS232-compatible at 300 baud. 6.7 REFERENCES Baker, R. W. , R. L. Whitney, and E. W. Hewson (1979): A low level wind measurement tech- nique for wind turbine generator siting. Wind Eng. 3:107-115. Baker, R. W. , R. L. Whitney, and E. W. Hewson (1979): Wind profile measurements using a tethered kite anemometer. Am. Wind Energy Conf., 16-18 April 1979, San Francisco, Calif., American Wind Energy Association, Washington, D.C. Delaurier, J. D. (1972): A stability analysis of cable-body systems totally immersed in a fluid stream. NASA CR-2021, Stanford Univ., Dept. of Aeronautics and Astronautics, Stanford, Calif., 93 pp. McMichael, J. M., and P. A. Klebanoff (1975): The dynamic response of helicoid anemometers. NBSIR 75-772, U.S. Dept. of Commerce, National Bureau of Standards, Washington, D.C, 54 pp. Shieh, C. F., and W. Frost (1980): Tether analysis for a kite anemometer. Conf. on Wind Characteristics and Wind Energy Siting, 19-21 June 1979, Portland, Ore., Pacific Northwest Laboratory, Richland, Washington. 37 7. REMOTE ACOUSTIC ELECTRONIC SOUNDING (RACES) P. Ravussin Federal Institute of Technology Lausanne, Switzerland 7.1 INTRODUCTION The Remote Acoustic Electronic Sounding system (RACES) can measure in the lower layer of the atmosphere the vertical profile of the three-dimensional wind vector, the vertical profile of temperature, and the vertical profile of humidity. The version de- scribed here measures only the vertical profiles of vertical wind and temperature. As the measurements are made at the same time and in the same sampling volume, the system can calculate in time the vertical profile of the thermal coefficient of turbulent diffusivity K h> K h = (7.1) 3T/9x 3 where T is the aleatory variable of the temperature, and T = T + T' , u 3 is the aleatory var- iable of the vertical component of the wind, and u 3 = u 3 + u 3 . T and u 3 are the statistical averages, defined in terms of probability density functions p(T) and p(u 3 ): co T = / T-p(T)dT (7.2) -273 and -(-CO / u 3 -p(u 3 )du 3 . (7.3) Since the RACES system measures time series, the ergodic assumption of the stationarity of the measured phenomenon must be made, to replace the statistical averages by time averages. The averaging time is a critical parameter which depends on atmospheric conditions and topography . 7.2 , OPERATION The RACES system is based on the physical properties of transmission and diffu- sion of sound in the atmosphere. Thus the basis of the instrument is an electro-acoustic device which transmits vertically in the atmosphere. 7.2.1 Sound Transmitter The sound transmitter (Fig. 7.1) contains an oscillator, which produces a sinus- oidal electric signal at a very stable frequency f = 1600 Hz, Af/f < 10~ 6 . The signal is transmitted through a switch (which is electronically operated) to a power amplifier and then to the electro-acoustic transducer. There the electric signal is transformed to an acoustic wave. The efficiency of the transducer is low: Ps H = — = 20% , (7.4) Po 38 pulse of sound acoustic enclosure power amplifier switch oscillator /. - transducer <<<*>> — . . . , , parabolic reflector Figure 7.1. Diagram of sound transmitter . where p = acoustic power, and p = electric power. (7.5) where u is the instantaneous voltage applied to the transducer and Z is the electrical impedance of the transducer at the oscillator frequency (1,600 Hz). The relation between the effective (u r c ) and peak (u ) values is given for a sinusoidal signal by ef f p 2 2 12 u ff = u = — u eft 2 p (7.6) In the low-power version used here p = 30 W, and the acoustic power is only 6 W. 7.2.2 Sampling Volume The sampling probe is the sound pulse itself. The sizes of the sampling volume are identical to the sizes of the sound pulse. The pulse duration : s t p = 50 ms. The length of the sampling volume is given by (7.7) I = C ' t s p C g is the speed of sound in the air, and varies slightly with the temperature according to the law C g = 20.05 /f (7.8) where T = temperature (K). At 20°C, £ - 17 m. The acoustic antenna is circular, which gives the pulse of sound a cylindrical form whose diameter varies with the altitude according to the law of diffraction of a circular opening. 9=1.22 A, where A = wavelength of the sound (m) , and d = diameter of the antenn; (7.9) 39 ///)// /?// //yy/y ////////////// pulse of sound Figure 7.2. Sound propagation from acoustic antenna. Figure 7.3. Acoustic intensity and power as a function of angle. A = r (7.10) For the 1600-Hz signal at 20°C, A = 0.21 m. For the present low-power version, the diameter of the antenna is 1.31 m. It is therefore possible to calculate at which altitude the diameter and the length of the pulse of sound will have the same value (Fig. 7.2): v 1 A 2 sin(1.22 -) d = 44 m (7.11) The intensity of the sound is, however, not constant in the sampling volume. It varies according to the law 1 = 1 2 J (| k sin 6) 2 o d rk sin B (7.12) where I = intensity at the center, J 1 = first-order Bessel function, k = wave number = 2n/\. Figure 7.3 shows that 80% of the power is emitted in an angle 6 = 0.5 6 . For the RACES system, the length and the diameter of the sound pulse reach the same size at an altitude of ~90 m. 40 Table 7.1. Corrections for relative humidity Temp. 100% H 0% H 0°C -0.23 +0.23 10°C -0.48 +0.48 20°C -0.93 +0.93 30°C -1.78 +1.78 7.2.3 Time Constant The time constant in the RACES system is the time that the sound takes to propa- gate along a distance equal to the length of the sound pulse. This is obviously the time duration of the pulse (50 ms). The RACES system is therefore capable of measuring the in- stantaneous value of the spatial average of the parameters T and u 3 in the sampling volume. , V T gv = f /// T dV (7.13) u. = - /// u- dV . (7.14) 3ev V 3 7.3 MEASUREMENT OF THE TEMPERATURE PROFILE The speed of sound propagation in air is directly connected to the thermodynami- cal properties of the atmosphere. Because of this the RACES system measures directly the absolute value of the temperature. 7.3.1 Theory The speed of sound in the air depends solely on the temperature. This can be deduced from the equation of the mechanics of fluids and the thermodynamic equations of the adiabatic processes, YKT (7.15) y where C s = speed of sound (m/s), y = ratio of the isobaric and isochoic specific heats, R = gas constant (J mole" 1 K _1 ) , |J = molecular weight (mole -1 ), and T = absolute temperature (K) . In a dry atmosphere the proportions of the main gases, Ar, N 2 , and 0g , are constant. In this case \J is also constant. y does not vary with atmospheric pressure but does, very slightly, with temperature. However, the effect is negligible within a 10°C temperature variation. 7.3.2 Effect of the Air Moisture Unfortunately the effect of the water vapor in the atmosphere is too strong to be negligible, especially at high temperatures. Because this version of the RACES system does not measure the vertical profile of humidity, the assumption of a constant 50% humidity profile was made. As seen in Table 7.1, the influence of moisture decreases rapidly with temperature. Table 7.1 gives the corrections for 100% and 0% relative humidity compared with 50% relative humidity, at different temperatures. 41 pulse of sound C s= 330 m/s C = 3 10 8 m/s UHF transmitting antennas galvanic screen UHF receiving \\y \L/ antennas array Figure 7.4. Measurement of speed of sound propagation in the atmosphere. 7.3.3 Effect of the Vertical Wind The measured speed of sound is the sum of the effective speed of the sound and the vertical wind. Since the vertical wind is measured at the same time and in the same sampling volume, the computer can calculate this correction. 20.05 /f + u. (7.16) 7.3.4 Tempe ra ture Me a s u reinen t Since the speed of sound in the atmosphere depends mainly on the temperature, the propagation speed of the sound pulse is directly measured with a CW Doppler radar. The principle is illustrated in Fig. 7.4. The wavelength of the radar must be exactly twice the wavelength of the sound pulse (which varies with the temperature). In order to improve the range, a feedback device was patented which corrects the frequency of the CW Doppler radar according to the frequency deviation of the return Doppler signal. The circuit used in this version is a phase-locked loop (PLL) . 42 Al the point of equilibrium the Doppler signal must have the same frequency as that of the sound pulse. Because of this, the same oscillator was used to produce the sound pulse and the reference signal for the feedback device. The relationship between the speed of sound and the radar frequency is given by the Dopp] er-Fizeau law: Av 2v (7.17) where Av = received Doppler-Fizeau frequency, V = frequency of the emitter, C„ = speed of sound, and C = speed of the electromagnetic wave in the atmosphere. Since C is about 10 6 times faster than C s , the speed of sound measurements made by RACES are essentially instan- taneous . The time interval between the starting of the sound pulse and the time of measure- ment gives the height of the measurement: t h = / C dt . (7.18) s t o Tiie system measures the ratio Av/v direct J y and is therefore independent of the phase effect introduced by the low-pass filter of the PLL. 7.3.5 Description of the System 7.3.5.1 U HF transmitter The UHF transmitter (Fig. 7.5) consists of a low-noise voltage-controlled oscil- lator (0.66 - 0.78 GHz) followed by a power amplifier (1 W) and a four-Yagi , 26-element, wide-band antenna system. The opening angle of the antenna is about 20°. 7.3.5.2 UHF receiver The UHF receiver consists of an array of five 26-element (15.5-dB Yagi) antennas, electronically connected to the receiver. The purpose of the array is to compensate for the effect of sound pulse displacement by the horizontal wind. The antennas are followed by a 54-dB low-noise UHF preamplifier and the Doppler mixer. For practical reasons only one antenna was used for the Boulder comparisons. 7.3.5.3 LF receiver and converter The low- frequency receiver consists of a low-noise preamplifier, a gyrator filter with adjustable Q factor from to 1,000, a linear amplifier (0 - 40 dB) , a log- amplifier, and a signal shaper circuit. The signal is then transmitted to an adjustable digital divider and to a UHF ratiometer with IEC-Bus interface. The measurement time is 12.5 ms . The measurement interval is 70 ins. This is because it takes about 35 ms to transmit the information to the computer through the IEC-Bus. Consequently, the tempera- ture is measured every 20 m only. 7.4 MEASUREMENT OF THE VERTICAL WIND PROFILE 7.4.1 Theory The sound pulse propagating vertically in the atmosphere is diffused by the small inhomogeneities of the air. The coefficient of diffusion oa, depends on the angle of diffusion of the thermal and wind turbulence according to the following expression: 43 Receiving antenna array UHFL> acoustic antenna transmitting antenna \ ^ to vertical wind / -\ measurement "Vl/ mixer (X, low noise amplifier logic circuit {>' Lt" power amplifier LF receiver and converter oscillator and PLL [>UHF power amplifier LF receiver and converter o UHF oscillator period meter ratio meter Temperature measurement Vertical wind measurement To mini- computer i o mini- computer Figure 7.5. Block diagram of RACES system. 1/3 a, = 0.03 k cos' i(== cos -^ + 0.13 ==) (sm -~r) v 2 z T 2 l (7.19) where k = wave number of the sound,

T 2 The sound is back-scattered only by the thermal inhomogeneities of the atmosphere. 44 7.4.2 Principle of the Wind Measurement The small-scale thermal inhomogeneities in the atmosphere have an average general movement u 3 in relation to the receiver's antenna. In this case, according to the Doppler law, the back-scattered sound that reaches the acoustic antennas has a frequency slightly different from that of the emitting antenna. 2f C s (7.20) where Af = Doppler frequency, f = frequency of the pulse of sound, u 3 = vertical wind, and C s = speed of the sound in the atmosphere. C s is measured by the temperature-measuring part of the RACES system. Because the Fourier transform needs a stationary signal, the frequency measure- ment of the received signal was not made with a spectrum analyzer. Instead the average period of the signal was directly measured with a zero-crossing technique. The average is taken on 20 periods of the signal. 7.4.3 LF Receiver and Converter The acoustic antenna signal is switched (a few milliseconds after the emission of the sound pulse) to a low-noise preamplifier, a gyrator filter with adjustable Q factor from to 1000, a linear amplifier (0 - 40 dB), a log amplifier, and a signal shaper cir- cuit. The signal is then transmitted to an adjustable digital divider and to a period- meter (interval timer) with an IEC-Bus interface. The minimum time interval of the period- meter is 0.1 (Js . The timing considerations are exactly the same as in the temperature measurements. Consequently the vertical wind is measured approximately every 10 m. 7 . 5 DATA ACQUISITION AND PROCESSING The RACES system uses a PDP11-03 16-bit minicomputer with 32-k-word RAM memory, display, an LA 35 printer, and an RX02 floppy disk driver with two single-side, double- density disk drivers. The operator can choose the time interval from (immediate) to 60 minutes, the number of soundings for averages from to 400, and whether to print all the data, the selected data, or the statistical data, or to keep the data on file in a second floppy disk. A sample of the output is shown in Fig. 7.6. 7 . 6 P URPOSE OF THE COMPARISONS Our purpose in BLIE will be to compare the measurements made by the RACES system with those made by the conventional tower instruments and by other remote sensing systems. It must, however, be considered that the RACES system has a sampling volume much larger and a sampling time much shorter than those of conventional instruments on the tower. The range of the RACES system used will be rather low (M50 m) because the high- level version is too bulky and too heavy to be transported easily overseas. Another reason for the low range will be that we will have only one UHF receiving antenna instead of five. 4S 300 SON! AflCC 04 -SEP-79 11 !40J00 HAUTEUR T - BAR SIGMA NB , BE MESURES (M) 394. (DEG-C) 29.98 (DEG-C) 0.00 VALIDES 0.3 (X) I Tl 368. 0.00 0.00 0.0 I 343. 25.32 1.64 0.7 I T 317. 24.76 0.44 0.7 I T 291 . 27.26 0.00 0.3 I T 1 265. 27.69 0.91 1 .3 I T I 239. 28.99 o 22 2.3 1 T I 213. 28.23 2.37 3.7 1 T I 186. 27.35 2 . 92 4.7 I T I 160. 27.95 2.42 5.3 I T I 134. 27.51 1 .93 10.7 r T I 108. 27.91 2.27 13.0 I T I 82. 28.09 1.70 24.3 i T I 56. 28.68 1 .53 61.3 I T I 30. 28.42 1 .32 95.7 I T I 300 SONIi AHCQ 04- -SEP-79 — ii Af\ * r\r\ RLlL O HAUTEUR U - BAR SIGMA NB BE MESURES (M) 385. (M/S) -0.44 (M/S) 0.02 VALIDES 1 .3 r/.) i u I 372. 0.00 0.00 0.0 I I 360. 0.00 0.00 0.0 I I 347. 0.00 0.00 0.0 I I 335. -0.46 0.00 1 .0 I Ul I 322. 0.00 0.00 0.0 I I 310. -0.17 0.11 2.0 I u I 298. 0.00 0.00 0.0 I I 285. -0.33 0.02 1 .0 I u I 273. -0.43 0.01 1 .3 I u I 260. 0.00 0.00 0.0 I I 248. 0.00 0.00 0.0 I I 235. -0.29 0.21 3.0 I u I 223. 0.00 0.00 0.0 I I 210. -0. 49 0.01 1 .3 I u I 197. -0.43 0.01 1 .3 I u I 184. -0.23 0. 16 2.7 I Ul I 171. 0.00 0.00 0.0 I. I 158. -0.35 0.12 1 .3 I w I 145. -0. 13 0. 13 3.0 I u I 132. -0.06 0.01 1 .3 i u I 119. 0.00 0.00 0.0 i I 106. 0.00 0.00 0.0 I I 93. 0.22 0.00 1.3 I ui I 80. 0.00 0.00 . 0.0 i. I 67. -0.01 0.01 1 .7 I u I 54. -0.24 0.01 1.7 i III I 41 . 0.08 0.02 1 .3 I Ul I 28. 0.33 0.02 1 .7 I U I 15. 0.00 0.00 0.0 i I Figure 7.6. Sample output sheet, 46 8. FM-CW RADAR R. B. Chadwick and K. P. Moran NOAA/ERL/Wave Propagation Laboratory Boulder, Colorado, U.S.A. 8. 1 INTRODUCTION The frequency-modulated, continuous-wave (FM-CW) radar first developed by Richter (1969) combines high sensitivity necessary for detection of clear-air echoes with ultra- high resolution (<1 meter) and virtual freedom from ground clutter, features which cannot be achieved in pulse radars used for monitoring atmospheric structures. Atlas et al. (1970), Gossard et al. (1970, 1971), and Bean et al. (1971) have described the use of this new tool in studying a variety of micrometeorological processes. Recently, however, FM-CW systems have enjoyed an added dimension: Doppler wind-measurement capability (e.g., Chad- wick et al., 1976a, b; Chadwick and Strauch, 1979). The system can be operated in two modes: for high-resolution studies of atmospheric structure, the radar is operated in the range-only mode (to measure reflectivity as a function of range); for wind profiling, the radar is operated in the range-Doppler mode. The two modes of operation use the same equip- ment, differing only in sweep rates and sampling schemes. In the range-only mode of operation, the antennas are usually pointed vertically, typically providing a maximum range of about 3 km or less in clear air, but much greater in the presence of targets such as hydrometeors , chaff, and insects. The WPL equipment pro- vides 500 range cells within this altitude, yielding cells and hence resolution about 6 m or less in range. The beamwidth is some 0.05 radians, so that the interrogated cells are generally shaped like thin discs. In the range-only mode of operation, the output displays show regions of enhanced atmospheric refractive-index fluctuations. The time history of these records reveals the advection of structures passing overhead within the PBL during the observation period, as well as non-stationarity in the PBL itself (associated, for example, with the rise of the convectively mixed layer during the morning hours). From the resulting data set it is very easy to discern layers of high refractive-index variability and the behavior of these layers during the day. In the range-Doppler mode the antenna can either be aligned in a given fixed direction or scanned in azimuth. The maximum range for clear-air measurements depends on the elevation angle. Looking vertically, the maximum range is about 3 km and this maximum range increases as the antenna beam is lowered toward the horizon. The number of range gates and the number of spectral points are variable, subject to the constraint that the product of range cells and the number of spectral points per range cell must equal 500, which is the number of points available at the output of the signal processor. Normally, ten range cells with 50 spectral points each are used, providing radial wind velocity measurements at ten equally spaced intervals out to the maximum range. Such a measurement yields only the radial component of the wind, i.e., the wind component parallel to the antenna beam direction. To derive profiles of the total vector wind, the airspeeds are measured with the radar looking in two or more directions. Horizontal homogeneity of the wind field within the scanning volume is assumed. Wind-velocity measurements made while the radar antenna is scanning azimuthally yield a so-called velocity-azimuth display (VAD) . Total wind profiles as well as convergence profiles and estimates of shearing and stretch- ing deformation can be obtained from a VAD scan. The time required for one vertical pro- file of wind speed and direction is 30 s. A large fraction of this time is expended in steering the antenna. Doppler sensing of the radial wind component profile is relatively rapid, typically once per second. A recent improvement is the capability to operate the radar at very low elevation angles. Before these improvements, the minimum elevation angle was about 30° from horizontal The increased return from the ground clutter at lower elevation angles caused saturation 47 S&JJl Figure 8.1. The transmitting and receiv- ing antennas of the WPL FM-CW radar, shown pointed vertically. The BAO 300-m tower is in the background. The trailer on the right houses the electronics for the radar systems. of the signal processor. Now the radar can operate at an elevation angle of 5°. The maximum range at these lower angles is greatly increased. In the summer-time clear air, ranges exceeding 10 km are possible at 8° elevation angle. For hydrometeor return, the maximum range can exceed 40 km. The radar has mapped thunderstorms at 40 km range. As indicated above, the optically clear air targets for the FM-CW radar are half- radar-wavelength Fourier components of fluctuations in refractive index associated with atmospheric turbulence. Of course the radar detects other targets, including hydrometeors , insects, clouds, aircraft, and balloons. As a rule these other targets produce radar echoes of sufficiently distinctive character that they are readily distinguishable from clear-air returns, so that no misinterpretations arise. Indeed, to the extent that insects and chaff follow the mean flow, they simply increase the signal-to-noise ratio and actually aid the wind measurement process. However, when the backscattered power exceeds a certain level, the signal processor saturates and quantitative wind and backscatter intensity measurements deteriorate in quality and reliability. .2 DETAILS OF THE WPL FM-CW RADAR The Wave Propagation Laboratory FM-CW radar is mobile and transported on two trailers. Figure 8.1 shows the radar receiving and transmitting antennas and their mount, with the BAO tower in the background. The radar transmitter, receiver, and data processing electronics are housed in a trailer that is not shown. The major parameters describing system performance are listed in Table 8.1. No maximum range is given since this depends upon atmospheric conditions and/or the availability of suitable targets, factors that vary diurnally and seasonally. Table 8.1. FM-CW radar performance parameters Average transmitted power 200 W Antenna diameter 2.44 m Wavelength 10 cm Receiver noise figure 2.2 dB Minimum range 15 m Minimum detectable signal -155 dBm Range resolution (adjustable) >1.65 m Velocity resolution (adjustable) >3 cm/s 48 During the intercomparison experiment, the FM-CW radar will be operated at an elevation angle of 60° with a maximum range less than 3 km. This will provide height coverage to about 2 km. Because of the requirement that wind profiles be available within 24 h after they are taken, the data will be reduced by hand. This means the radar will be operated only during selected data-taking intervals, mostly during the day. Also, this will preclude obtaining winds from the lowest range bin where special processing techniques are needed to determine the sign of the Doppler velocity. To facilitate comparisons, the radar will be pointed either to the west or to the south during data-taking periods. While attempts will be made to measure both components for a few 20-min periods, the normal mode of operation will be to measure only one compo- nent during one 20-min period. The hand processing algorithm that determines which bin of the velocity spectrum contains the peak assumes that the peak represents the mean value. This assumption could introduce errors beyond those normally expected in an intercomparison such as this. The first type of error is that due to simple human error in locating the spectral peak. This would normally be a large error. In some instances these points would be isolated; in other instances, the points in error may occur in sequences. Careful processing should minimize the occurrence of such errors. The second type of error is due to the discrete nature of the output velocity spectra and the fact that only 50 velocity values can be selected over the range of +10.5 m/s. This "discreteness" introduces errors in the range of +0.5 m/s. A third type of error is that caused by using the peak of a non-symmetric spectrum as the mean. The size of this error cannot be estimated without having some measure of the spectral asymmetry. The radial velocity spectra taken by the FM-CW radar are on file and are avail- able to any BLIE participants. 8.3 REFERENCES Atlas, D., J. I. Metcalf, J. H. Richter, and E. E. Gossard (1970): The birth of "CAT" and microscale turbulence. J. Atmos. Sci. 27:903-913. Bean, B. R., R. E. McGavin, R. B. Chadwick, and B. D. Warner (1971): Preliminary results of utilizing the high resolution FM radar as a boundary layer probe. Boundary Layer Meteorol. 1:466-473. Chadwick, R. B., K. P. Moran, R. G. Strauch, G. E. Morrison, and W. C. Campbell (1976a): Microwave radar wind measurements in the clear air. Radio Sci. 11:795-802. Chadwick, R. B., K. P. Moran, R. G. Strauch, G. E. Morrison, and W. C. Campbell (1976b): A new radar for measuring winds. Bull. Am. Meteorol. Soc. 57:1120-1125. Chadwick, R. B., K. P. Moran, G. E. Morrison, and W. C. Campbell (1978): Measurements showing the feasibility for radar detection of hazardous wind shear at airports. Technical Report AFGL-TR-78-0160 , Air Force Geophysical Laboratories, Hanscom Air Force Base, Bedford, Mass. Chadwick, R. B., and R. G. Strauch (1979): Processing of FM-CW Doppler radar signals from distributed targets. IEEE Trans. Aerosp. Electron. Syst. AES-15 : 185-188 . Gossard, E. E., J. H. Richter, and D. Atlas (1970): Internal waves in the atmosphere from high-resolution radar measurements. J. Geophys . Res. 75:3523-3536. Gossard, E. E., D. R. Jensen, and J. H. Richter (1971): An analytical study of tropo- spheric structure as seen by high-resolution radar. J. Atmos. Sci. 28:794-807. Richter, J. H. (1969): High resolution tropospheric radar sounder. Radio Sci. 4:1261-1268. 49 9. DUAL-DOPPLER RADAR R. A. Kropfli NOAA/ERL/Wave Propagation Laboratory Boulder, Colorado 9.1 INTRODUCTION Although the Wave Propagation Laboratory developed its X-band dual-Doppler radars primarily to study motion fields within precipitating clouds (e.g., Miller and Strauch, 1974; Miller et al., 1975; Kropfli and Miller, 1976; Dye et al., 1978), the same radars have also been applied in PBL studies, both in small experiments (e.g., Wilson, 1970; Frisch and Clifford, 1974; Gossard and Frisch, 1976), and as part of large field programs such as METROMEX (e.g., Kropfli and Kohn, 1978). In such studies the radars use either hydrometeors , such as snowflakes (Wilson, 1970), or artificial chaff (e.g., Gossard and Frisch, 1976; Kropfli and Kohn, 1978) as tracers, deducing wind velocities from the Doppler shifts measured in the echoes from these targets. Scanning the radars through large vol- umes has provided a tremendous step forward in our visualization of boundary-layer flow fields . Despite the wide variety of field programs in which these radars have been used, there has not been an opportunity until now to make detailed comparisons of the Doppler radar wind fields with data from other remote sensors or in-situ instruments. An experi- ment called PHOENIX provided this opportunity in September of 1978, and the first results of this experiment are presented here. The focus of this experiment was a 300-m instru- mented tower, the Boulder Atmospheric Observatory (BAO) (Kaimal, 1978). The instruments on this tower, along with other remote, ground-based, and aircraft-borne sensors, were used in these intercomparisons . One of the many components of PHOENIX was an array of three Doppler radars: two NOAA/WPL X-band radars, and an NCAR C-band (CP-4) radar. Analyses described here involve only the two X-band radars. These radars were located to optimize observations near the 300-m-high, instrumented BAO tower. Short (^15-km) radar baselines were chosen to optimize spatial resolution, a luxury not possible in past multiple Doppler radar experiments. In addition to these intercomparisons, an important goal of PHOENIX was to im- prove our understanding of physical processes in the PBL. Understanding of a physical process almost always follows our ability to observe and measure that process in a better way. We are therefore hopeful that the first PBL flow visualizations presented here, and the ones that will be produced later, will be followed by a corresponding increase in our understanding of the PBL. 9.2 DESCRIPTION OF THE EXPERIMENT Since the backscattered signal from the convective PBL is usually too weak to be observed reliably by the radars involved in PHOENIX, chaff must be dispensed from an air- craft over an area of several hundred square kilometers. X-band chaff was distributed for several hours at a time along 15-km crosswind flight legs. Usually, the flight patterns were adjusted to dispense chaff about 30 to 45 min upwind of the target area. When the winds were weak and variable in direction, a zig-zag pattern covering an appropriate box was chosen. Convective activity was usually sufficiently strong between 1100 and 1800 (all times are given in MDT) each day to disperse the chaff uniformly throughout the convective PBL in the test area. Radar echoes from the resulting chaff cloud were usually greater than 10 dB above noise power within most of the region of interest. The two identical N0AA systems were operating at a wavelength of 3.22 cm, peak transmitted power of 20 kW, pulse width of 1.0 fjs, and beamwidth of 0.8°. 50 Longmont ^ Doppler Radar T BAO Tower ▲ Optical Triangle Sites +NOAA Radc 26 27 23 24 2 19 20 21 13 14 15 16 12T 7 8 9 1<^ 3 4 5 1 2 "V 22 11 ®n\ + NOAA Radar — 0- 5 Kilometers Figure 9.1. Positions of radars and optical triangle relative to BAO tower. PAM stations are indicated by numbers. Figure 9.1 shows the location of the radars relative to the tower. The NCAR portable automated mesonetwork (PAM) and the NOAA optical triangle are also shown. The 13- to 18-km separation between radars is much smaller than is normally used in multiple Dop- pler radar experiments. Thus, the radars were able to scan the entire depth of the PBL (^-2 km) over an area of several hundred square kilometers while observing air motions at wavelengths as small as 600 or 700 m. Volume scans were completed in less than 90 s. Siting the radars equidistant from the tower also had the important advantage of equalizing the radial and tangential dimensions of the three radar pulse volumes to about 150 m at the BAO. 51 Table 9.1. Summary of radar scan characteristics for PHOENIX Scan number/name Volume time (s) Cartesian grid element AX, AY, AZ (km) Sample density number (km 3 ) Half-amplitude wavelength after filtering (km) Area covered (km 2 ) 100/Standard 72 * j. ' ~~ l I ".- - » •* ^ i i — i — r 5 - -3 -2.5 -2.0-1.5 -1.0 -5 .00 10 1.5 20 25 3.0 35 40 -3.0-2.5-2 0-1.5-1.0 Distance East of Tower (km) _i i i i ' i .00 .5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Figure 9.5. Vertical (XZ) sections through the wind field at 1531 (left) and 1533 (right) MDT on 21 September 1978 for Y=0 . 5 km. (a) -1.0 -3.0 -2.0 * 3.0 2.0 1.0 ■1.0 10 2.0 3.0 U Radar (m/s) •100 Series *200 Series o300 Series — -2.0 -3.0 (b) —3.0 o o +•1.0 o -3.0 -2.0 -1.0 ■2.0 o -1.0 10 2.0 3.0 W Radar (m/s) • 100 Series *200 Series o300 Series -2.0 -3.0 yi 1.0 2.0 3.0 » V Radar (m/s) • 100 Series *200 Series o300 Series Figure 9.6. Scatter plots of the components from the tower at the 300-m level with components from the dual-Doppler radar analysis : (a) U component; (b) W component; (c) V component. 54 grid volume. The two methods generally agree to within 0.5 m/s despite the volume aver- aging by the radar, as opposed to the line average represented by the tower values. There is a slight underestimation in the magnitudes of the three components, which could be the result of this volume averaging and ground-clutter biasing. Whenever comparisons of radar- and aircraft-derived fields are made two problems should be considered to ensure accurate and valid comparisons. First, the aircraft must be positioned accurately in time and space with respect to the radar data. Spatial agreement within at least 0.2 km and temporal agreement within 1 min are needed for the assumption of stationarity of the turbulent wind fields to be valid. For differences much larger than about 0.2 km and 1 min, changes can occur that seriously degrade the comparison of radar and aircraft data. Although the inertial navigation system of the aircraft can drift significantly, visual fixes were used in this experiment to correct the aircraft locations to within 0.2 km. Even though every effort was taken to produce simultaneous measurements, some data comparisons had to be made with aircraft and radar data separated by as much as 2 to 3 min. The second problem to be considered is that of filtering aircraft and radar data such that the spectral content is as similar as possible. The aircraft velocity data are collected at a rate of about 20 Hz, with the samples being essentially independent. The radars collected radial velocity data that were interpolated to Cartesian grids having grid elements 200 or 250 m on a side. These radar data were processed as described in the preceding section. In order to match the radar-derived scales, a Gaussian filter with a half width of 0.2-km was applied to the aircraft data. Sample comparisons of multiple Doppler radar data and aircraft data are presented in Figs. 9.7 and 9.8. Single data points have been added in Fig. 9.7 to indicate BAO tower wind values (The tower is 0.75 km north of the east-west flight track.) These measurements and others like them indicate agreement between multiple Doppler radar, aircraft, and tower measurements to better than 1 m/s in most cases. o > -2-10 1 2 Distance East of Tower (km) Figure 9.7. Compari ponents (U, V, and Doppler data (at 1 aircraft data (at the same day. Sol sent aircraft data represent radar da 0.15, 0.10, and 0. aircraft, and inve tively. BAO tower dicated by * (at D (at 0. 15 km) . son of wind com- W) derived from 100 hours) and 1108 to 1110) on id lines repre- ; dashed lines ta . Heights are 50 km for radar, rsion respec- data are in- . 1 km) and 55 E >> _o 11 > (v)l 1 \ • \ J A \ / v V -2-1 1 2 Distance East of Tower (km) Figure 9.8. Comparison of wind com- ponents derived from Doppler data (at 1335 hours) and aircraft data (at 1336 to 1338) as in Fig. 9.7. Heights are 0.5, 0.6, and 0.8 km for radar, aircraft, and inver- sion, respectively. 9.4 FLOW VISUALIZATION IN THE PLANETARY BOUNDARY LAYER Figure 9.9 is an example of how the use of the various scan types summarized in Table 9.1 allows us to see the PBL motions with different magnifications. The figure contains a large-scale 600 series scan, a 200 series scan, and a high-resolution 300 series scan, all taken within 8 min. Boxes have been drawn over the 600 series grid to indicate where the 200 and 300 series grids lie. The same features seen in the 600 series grid can be seen with increased detail in the 200 series. The 600 series display shows a very chaotic wind field with sharp wind gradients aligned roughly along the mean wind direction, which was 2.8 m/s from the SSE. Scales of motion here are about 3 km or about three times the depth of the convective boundary layer at this time. An example of an unusual feature, suggestive of flow around a jet, is depicted in the 200 series and 300 series displays of Fig. 9.10. An updraft of about 1.5 m/s was observed by radar at the location (0.5, 0.8) at a height of 0.825 km. Such a weak updraft would not be expected to act as a barrier to the prevailing flow. This feature was clearly evident in the wind patterns for about 10 min and was observed to track with the mean wind. Recognizable features in the eddy field could usually be tracked along the mean wind for about 10 or 15 min. 56 63 i 1 ! 1 1 .03 .■'■\ ". • * * .43 • • • '/,y'. . , * i ' - '. ■ .*»•»■• ■ ' ' 18 • ' •' ' ' : ' , r> , , - • ' ." • . . , , W " ' ^ ? ^ ' ' '* u ■ . . '4 * 4 i ~ , 9 ' 78 ■ ' t » 38 1 < * s 1 1 1 1 300 Series High Resolution Z = 0075 km T = 1522 2.0 m/s -138 -0.78 -018 043 1.03 1.63 200 Series Medium Resolution Z = 0100 km T= 1517 2.0 m/s -2.0 ^ %.-}&'>' -Hi- • ipiSSW^ ■ 4 II T y//, «r^ 600 Series Low Resolution Z = 150 km T = 1525 2.0 m/s -4.0 -2.0 Figure 9.9. Horizontal eddy fields at high, medium, and low resolutions -2-10123 km East of BAO -1.0 -0.5 0.5 1.0 km East of BAO Figure 9.10. Horizontal eddy fields on 21 September 1978: (left) at Z=0.9 km, obtained from 200 series scan at 1538; (right) at Z=0.825, obtained from 300 series scan at 1540. r )7 9.5 SUMMARY We have presented a brief description of the dual-Doppler technique for measuring three-dimensional wind fields, intercomparisons with other in-situ measurements of the three wind components, and a sample of the flow fields obtained during the recent PHOENIX experiment at the BAO. Although the analysis of these data is far from complete, we expect these data and similar data sets to have important impacts on remote sensor techniques and also on our understanding of the dynamics of the PBL . 9.6 REFERENCES Dye, J. E. , L. J. Miller, B. E. Martner, and Z. Levin (1978): Growth and recirculation of precipitation in an evolving convective storm. Prepr. Conf. on Cloud Phys . and Atmos . Electr., 31 July-4 August 1978, Issaquah, Washington, American Meteorological Soci- ety, Boston, Mass., pp. 528-533. Frisch, A. S., and S. F. Clifford (1974): A study of convection capped by a stable layer using Doppler radar and acoustic echo sounders. J. Atmos. Sci. 31:1622-1628. Gossard, E. E., and A. S. Frisch (1976): Kinematic models of a dry convective boundary layer compared with dual-Doppler radar observations of wind fields. Boundary Layer Meteorol. 10:311-330. Kaimal, J. C. (1978): NOAA instrumentation at the Boulder Atmospheric Observatory. Prepr. 4th Symp . Meteorol. Obs . and Instrum., 10-14 April 1978, Denver, Colorado, American Meteorological Society, Boston, Mass., pp. 35-40. Kropfli, R. A., and N. M. Kohn (1978): Persistent horizontal rolls in the urban mixed layer as revealed by dual-Doppler radar. J. Appl. Meteorol. 17:669-676. Kropfli, R. A., and L. J. Miller (1976): Kinematic structure and flux quantities in a convective storm from dual-Doppler radar observation. J. Atmos. Sci. 33:520-529. Miller, L. J., and R. G. Strauch (1974): A dual-Doppler radar method for the determination of wind velocities within precipitating weather systems. Remote Sensing Env. 3:219- 235. Miller, L. J., J. D. Marwitz, and J. C. Fankhauser (1975): Kinematic structure of a Colorado thunderstorm. Prepr. 16th Radar Meteorol. Conf., 22-24 April, Houston, Texas, American Meteorological Society, Boston, Mass., pp. 128-133. Rummler, W. 0. (1968): Two pulse spectral measurements. Tech. Memo. MM-68-4121, Bell Telephone Laboratories, Whippany, N.J. Wilson, D. A. (1970): Doppler radar studies of boundary layer word profile and turbulence in snow conditions. Prepr. 14th Radar Meteorol. Conf., Tucson, Arizona, American Meteorological Society, Boston, Mass., pp. 191-196. 58 10. REMOTE SENSING OF TEMPERATURE PROFILES WITH COMBINED ACTIVE AND PASSIVE SENSORS M. T. Decker NOAA/ERL/Wave Propagation Laboratory Boulder, Colorado, U.S.A. 10. 1 INTRODUCTION Project PHOENIX (Hooke, 1979), carried out at the Boulder Atmospheric Observa- tory (BAO) during September 1978, involved a variety of atmospheric sensors including aircraft, the NCAR PAM network, radars, lidar, acoustic sounders, microwave radiometers, optical wind sensors, radiosondes and fixed level balloons, and the 300-m BAO instrumented tower. The many goals of this project included evaluation and comparison of various remote sensing systems. Among these was the comparison of atmospheric temperature profiles ob- tained from microwave radiometers with profiles from standard tower and radiosonde sensors, and especially the usefulness of information from active sensors such as microwave radars and acoustic sounders in improving the resolution of vertical structure in the radiometric temperature profiles. It will be shown that the active sensor information can indeed be useful but that questions remain regarding the proper interpretation of the observed echoes . 10.2 INSTRUMENTATION The measurements reported here were made by microwave radiometer, FM-CW radar, and radiosondes colocated at the BAO site. Radiosonde equipment was a standard GMD system operated by the NCAR Field Observing Facility, and 38 flights were made during the month . Three microwave radiometer systems were operated at the BAO site during the PHOENIX experiment. Data reported here are from the Scanning Microwave Spectrometer (SCAMS) operated by personnel from the Jet Propulsion Laboratory. This is a 5-channel instrument similar to that flown aboard the Nimbus 6 satellite. It has one frequency at the water vapor absorption line at 22.235 GHz, three frequencies (52.85, 53.85, and 55.45 GHz) in the oxygen absorption complex, and a frequency of 31.65 GHz in the window between these absorption bands. The instrument scans in a vertical plane from a zenith angle of 58.3°, through the zenith, and to 28.1° on the other side of zenith. The scan steps in 7.2° increments with a dwell time of about 1 s at each step. Two additional steps are used to point the antennas at calibration targets, and the entire sequence is repeated approximately once each 16 s. The sky radiation measurements for all channels at zenith and for 55.45 GHz at 58.3° zenith angle were used to retrieve the temperature profiles reported here. The 1-s measurements (at 16-s intervals) were averaged over a period of about 7 min before being used in the profile retrieval algorithm. The FM-CW radar was operated during PHOENIX by the NOAA Wave Propagation Labora- tory. This radar (Chadwick et al., 1976) operates at a wavelength of 10 cm with an aver- age transmitted power of 200 W. The 2.44-m transmit and receive antennas are steerable in elevation and azimuth. The radar operates in either of two modes: a high-resolution range-only mode or a range-Doppler mode with a wind-measurement capability. In the meas- urements reported here the radar operated in the range-only mode with the antennas pointed at the zenith. The high-sensitivity, high-resolution, low-ground-clutter qualities of this radar allow detection of the detailed structure of atmospheric refractive-index fluctuations in the lower atmosphere. Minimum range is 15 m, and in clear air the typical maximum range is 3 km. The presence of persistent layer echoes is used here as evidence of thermal structure, specifically an elevated temperature inversion. 59 CD X 3.0 2.5 2 1 5 1 0.5 Project Phoenix 11 Sept 1978 Radiometer 10 10 20 Temperature ( C) 30 CD I 30 2 5 2.0 1 5 1.0 0.5 Project Phoeni 7 Sept 1978 • Radiometer 10 10 20 Temperature ( C) 30 Figure 10.1. Comparison of radiosonde and radiometer temperature profiles for a case with simple vertical structure . Figure 10.2. Comparison of radiosonde and radiometer temperature profiles for the case of a ground-based inver- sion. 10.3 TEMPERATURE PROFILE RETRIEVAL Statistical retrieval algorithms (Waters et al., 1975; Westwater et al., 1975) are used to extract temperature profiles from the radiation measurements. In these algo- rithms we use a nine-element data vector consisting of the six radiation measurements as well as surface temperature, pressure, and relative humidity to obtain a minimum variance estimate of the temperature at any level. The available measurement frequencies also allow us to correct for the effect of radiation from clouds as described by Westwater et al. (1976). Examples of the effectiveness of this cloud correction technique are con- tained in a series of measurements reported by Decker et al. (1978). It has been further demonstrated by Westwater (1978) that if the presence and height of an elevated tempera- ture inversion can be observed, the retrieval algorithm may be derived from a statistical ensemble of atmospheres all of which contain temperature inversions at this height (or realistically, within some representative height range). This method of conditional statistics has been applied to a number of cases from PHOENIX, and examples are shown here, 10.4 RESULTS A sample comparison of temperature profiles from the radiometer and radiosonde for a case with little vertical structure is shown in Fig. 10.1. In the example of Fig. 10.2 the profile shows a ground-based temperature inversion. In cases such as these the radiometer profile is generally in good agreement with the radiosonde profile. An example of an elevated temperature inversion is shown in Fig. 10.3. In this case the structure of the profile is smoothed by the radiometer to the extent that the temperature inversion is not observed. It is in this type of profile that knowledge of inversion height would be a very useful piece of information. The FM-CW radar record at this time shows a persistent echo with maximum intensity at 507 m above the surface. The profile retrieval algorithm is 60 3.0 2 5 2 E * 1.5 ~E 0) I 1 Ob Project Phoenix 18 Sept 1978 -10 10 20 30 Temperature ( C 'C) 3 2.5- 2.0 E r 15 X 1.0- 0.5- \" ' I I Project Phoenix \\ \\ 18 Sept 1978 \ \ \ \ \\ Radiometer y, ,With Height \ \y Information \ V _ \ \ \ \\ \ Radiosonde Is I >7 I v \\ A i\ i 10 10 20 Temperature ( : 'C) 30 Figure 10.3. Comparison of radiosonde and radiometer temperature profiles for the case of an elevated inversion, Figure 10.4. Comparison of radiosonde and radiometer temperature profiles for the case of an elevated inver- sion, in which the radiometer profile is retrieved with knowledge of the height of the temperature inversion derived from radar echoes. then derived from a statistical ensemble, each member of which contains an elevated inver- sion with base in the height range from 400 to 600 m above the surface at Denver, Colorado. The profile retrieval resulting from the use of this algorithm is shown in Fig. 10.4. This profile is an improved representation of the radiosonde profile, and it is evident that the inversion height information has been helpful. It should be noted that the 200-m interval was used in the statistical ensemble so that enough profiles could be found in our data base to give a representative sample. A larger data base would allow this interval to be narrowed and presumably improve the retrieved radiometer profile. It must be pointed out that the above procedure required the assumption that the echo observed by the FM-CW radar was associated with an elevated temperature inversion. Such, of course, is not always the case; in fact, at the time of the profile of Fig. 10.2 the radar was observing an echo at a height of 157 m. If it is assumed that this echo is associated with an elevated inversion rather than the ground-based inversion, and the tem- perature retrieval is performed using conditional statistics with inversions between 100 and 300 m, the resulting profile is as shown in Fig. 10.5. It is obvious (when Fig. 10.5 is compared with Fig. 10.2) that this procedure has degraded the radiometrically retrieved profile. In view of a number of examples such as this observed during PHOENIX, additional work must be done to assure proper use of the echo height data. Methods for characterizing the echo data are being studied. A more basic study of the relation between the radio re- fractive index structure parameter which is observed by the radar and the profile or gra- dient of refractive index which is related to temperature and water vapor is being pursued. It is expected that the combination of active and passive sensors will result in improved remote sensing of profiles for research and operational use. 61 30 25 2.0 E *: r 1.5 .c D) I 1.0 05 0l_ -10 Project Phoenix 7 Sept 1978 Radiometer With Height Information Radiosonde 10 20 Temperature (°C) Figure 10.5. Comparison of radiosonde pro- file from Fig. 2 with radiometer profile derived with incorrect use of radar echo i nformation. 30 10.5 ACKNOWLEDGMENTS The radiometric measurements used here were made under the direction of Bruce L. Gary of the Jet Propulsion Laboratory, Pasadena, California. Russell B. Chadwick directed the work with FM-CW radar; Ed R. Westwater developed the retrieval algorithms for the radi- ometer data. Both are with the N0AA/ERL Wave Propagation Laboratory. 10.6 REFERENCES Chadwick, R. B., K. P. Moran, R. G. Strauch, G. E. Morrison, and W. C. Campbell (1976): Microwave radar wind measurements in the clear air. Radio Sci. 11:795-802. Decker, M. T., E. R. Westwater, and F. 0. Guiraud (1978): Experimental evaluation of ground-based microwave radiometric sensing of atmospheric temperature and water vapor profiles. J. Appl. Meteorol. 17:1788-1795. Hooke, W. H. (ed.) (1979): Project PHOENIX: The September 1978 Field Operation . NOAA/ NCAR Boulder Atmospheric Observatory Rept . No. 1, available from NOAA/ERL, Boulder, Colo. 80303, and from NCAR Publications Office, Boulder, Colo. 80307. Waters, J. W. , K. F. Kunzi, R. L. Pettyjohn, R. K. L. Poon, and D. H. Staelin (1975): Remote sensing of atmospheric temperature profiles with the Nimbus 5 microwave spec- trometer. J. Atmos. Sci. 32:1953-1959. Westwater, E. R. , J. B. Snider, and A. V. Carlson (1975): Experimental determination of temperature profiles by ground-based radiometry. J. Appl. Meteorol. 14:524-539. Westwater, E. R., M. T. Decker, and F. 0. Guiraud (1976): Feasibility of atmospheric temperature sensing from ocean data buoys by microwave radiometry. NOAA Tech. Rept. ERL 375-WPL 48, NOAA/ERL, Boulder, Colo. [NTIS No. 262-421]. Westwater, E. R. (1978): Improved determination of vertical temperature profiles of the atmosphere by a combination of radiometric and active ground-based remote sensors. 4th Symp . on Meteorol. Obs. and Instrum. , 10-14 April 1978, Denver, Colo., American Meteorological Society, Boston, Mass., pp. 153-157. 62 1 1 . WPL DOPPLER SOUNDER W. D. Neff, H. E. Ramm," and C. Wendt NOAA/ERL/Wave Propagation Laboratory Boulder, Colorado, U.S.A. 11. 1 INTRODUCTION This paper describes the use of the "complex covariance" frequency estimation technique in a microprocessor-controlled acoustic sounding system. A bistatic scattering arrangement was used during BLIE with fan-beam transmitters, a central receiver, and two orthogonal 300-m baselines. A frequency of 1250 Hz with a 100-ms pulse of 300 electrical watts was implemented. The Wave Propagation Laboratory has developed and tested a number of Doppler acoustic sounders during the past ten years. These used a variety of frequency estimation techniques and transmitter-receiver configurations. Direct spectral calculation, analog tracking devices (Kaimal and Haugen, 1977), and adaptive filter techniques were utilized. However, all these techniques required either expensive microcomputers or hardware to implement. With the development of microprocessors capable of using higher-level languages such as Fortran, an effort began in this laboratory to reduce the complexity and expense of Doppler sounding systems. Owens (1977) simplified the acoustic sounder electronics to a single printed circuit board and examined a simple frequency estimation technique referred to as "real covariance" for possible implementation with an LSI-11 microprocessor. The simplicity of the technique and the hardware developed by Owens led to more extensive field tests during the September 1978 Project PHOENIX experiment (Neff and Brown, 1979). However, these comparisons with tower data showed a systematic bias, leading to further laboratory testing in early 1979. Both real and complex covariance techniques were analyzed for the effect of white noise on the mean and variance of the frequency estimates. This analysis led to the choice of complex covariance as the preferred technique and the basis for the system de- scribed in this paper. An outline of the system to be described in the following sections is shown schematically in Fig. 11.1. 11.2 SYSTEM DESIGN AND DOPPLER ALGORITHMS The hardware for this system, with slight modification, was developed by E. J. Owens of WPL (Owens, 1977; also E. J. Owens, NOAA/ERL, Boulder, Colo., personal communica- tion) . Owens developed separate printed circuit boards for the acoustic sounder electron- ics and for the heterodyning and filtering of the signals required for the real covariance frequency estimation technique. Corresponding software was written for the 1978 Project PHOENIX experiment (Neff and Brown, 1979). The basis for the real covariance technique is the following algorithm (Owens, 1977): N-l I A. k Af = -±- cos" 1 -\ ^ -f , (111) 2TTT N-l N c u J I A. A. 1=1 X X '-'•'NOAA Commissioned Officer assigned to WPL. 63 where t s is the sampling period, f c the heteordyned carrier frequency, N the number of samples, and A ± the discrete sample of the signal. Sampling frequencies range from 500 to 1000 Hz, corresponding to reduced center frequencies of 125 or 250 Hz. The large number of samples results in either large memory requirements or a reduced number of samples per range gate to allow time for processing. The complex covariance technique is based on the following algorithm (Sirmans and iumgarner , 1975) : .\ (Q i+ i ^"QiW Af = wr tan i \ ( ^i Vi + h W 1=1 (11.2) 2000-Hz Tone Burst 11 Received Signal 1250-Hz Tone Burst Echosounder and Doppler Board PA PA Switch PA Array Leg-Flag lonostatic Bistatic Noise- Quadrature In-Phase LSI -11 Microprocessor Doppler Signals DecWriter Floppy Disk Fax Recorder Figure 11.1. Block diagram of LSI-11-controlled Doppler sounder. 64 where the complex time signal Z.(= Ii + iQi) is given by Z. = A(t.) cos t. 2oj + i A(t.) sin t. 2oj 11 10 1 1 o In this application, after the received signal is filtered (with a 300-Hz bandwidth) it is multiplied first by the carrier and then separately by the carrier phase-shifted 90 de- grees, providing the in-phase and quadrature components required in (11.2). After mixing, a low-pass filter (0 + 200 Hz at the -3 dB point) provides the final processing of the signals going to the computer, where the Nyquist frequency must be near the half-power point in the bandpass of these filters to avoid aliasing by the noise (R. J. Keeler, N0AA/ ERL, Boulder, Colo., personal communication). First-moment spectral estimators do not provide any information as to the noise content of the spectrum. A variety of techniques can be used to provide an approximate estimate of the signal-to-noise ratio. In our case we filtered and detected the noise below 1 kHz. Since normal background noise falls off with increasing frequency, we ad- justed the gain of the noise circuit to match the output of the signal circuit with the transmitters shut off. By using a broadband filter for the noise, we eliminated Doppler shifts calculated from nonwhite noise transients. Under most conditions, we observed that the variations in noise estimate between these two techniques were about 25 percent. A bistatic sounding arrangement similar to that described by Kaimal and Haugen (1977) was utilized. The fan-beam transmitters operated at 1250 Hz. The central receiver normally also acts as a monostatic transmitter at 2000 Hz. These choices of frequencies appear to avoid aliasing problems in the processing of the data. At present we do not calculate the vertical velocity, but rather assume that it averages to zero. 11.3 DIGITAL PROCESSING The details of the microprocessor system and its interfacing with an acoustic sounder have been provided by Owens (1977). To process complex covariance data we designed a general purpose program to sample a variety of acoustic data as well as the in-phase and quadrature components for use in equation (11.2) at 200 Hz and several additional channels at 100 Hz. These latter channels were used for the monostatic and bistatic intensities as well as the noise channel. An assembly language program is called shortly after the trans- mit gate. After the maximum range gate is reached, control is returned to the Fortran main program, signal-to-noise tests applied, and frequency shifts calculated and accumulated for a specified number of pulse repetition periods. For a maximum range of 600 m, a pulse repetition period of 5 s is used, with the last second dedicated to the Fortran processing. Following the required averaging period (normally 18 min) the wind components, speed, and azimuth are obtained from the individual components. These data, together with the signal intensities, noise level, and number of samples retained for each range gate, are printed out and also recorded on floppy disk for later analysis. A line-printer profile of wind speed and direction is also obtained for a quick visual impression of the data as shown in Fig. 11.2. 11.4 SIGNAL-TO-NOISE TESTS The hardware described in Section 11.2 was first tested in the laboratory by use of a signal from a waveform generator mixed with the output from a white noise source. These measurements were designed to test the relative merits of the two techniques. (The usual method of evaluation (e.g., Simians and Bumgarner, 1975) is to define a Gaussian spectrum, add a noise spectrum, and then perform an inverse Fourier transform.) Signal and noise levels were measured for reference purposes with two identical receivers set to the carrier frequency of 1250 "Hz with a centered bandwidth of 300 Hz. These mixed signals then served as inputs to the real and complex covariance pre-processor boards. Fifty 100- ms samples were then obtained at frequency shifts of 0, 10, 30, and 60 Hz. Mean and standard deviations were obtained by using the resulting frequency estimates. The results are shown in Figs. 11.3 and 11.4. 65 ACOUSTIC DOPPLER DATA AVERAGED FOR 20.0 MIN. STARTING TIME WIND SPEED AND DIRECTION MONTH 6 DAY 21 PROFILES HOUR 3 MIN 41 WIND HEIGHT DIRECTION (D) (M> SPEED DI SECTION 180 270 000 090 180 (M/S) (DEG FM N) +^^^^^^#^*^^^]|C3iC^^«+#3|C^^#^^}|£«3R^^^^^^^+^#^#«#^^^^^«^^«^#+«#^^«^#«3|C«^^NC^3|C^3K-t- 4 70. 6.4 205. * D s * 450. 7.0 207. * D S * 430. 8.0 209 . * D S * 410. 8.3 209. * n S * 390. 8.3 207. * D S * 370. 7.2 208. * n S * 350. 5.9 210. * n s * 330. 4.2 209. * D s * 310. 2.7 202. * D S * 290. 1.5 189. + D S + 270. 0.9 171. * S D * 2S0. 0.5 118. *S D * 230. 0.7 32. * S D * 210. 1.2 J 3. * S D * 190. 1.3 11. * S D * 170. 1.3 36. * s D * 150. 1.2 41 . * s D * 130. 1.4 41 . * s D * 110. 1.4 38. * s D * 90. 1.0 24. + s n + 70. 1.3 57. * s D * 50. 2.4 73. * s D * +***+***+***+***+***+***+***+***+***+***+***+***+***+***+***+ • J 10 WIND SPEED 15 (M7S) Figure 11.2. Sample output. Wind speed (S) uses scale along bottom of output in 0.25-m/s intervals. Wind direction (D) uses scale along top of graph in 5° intervals. 11.4.1 Conclusions 11.4.1.1 Real covariance The estimates described above were biased by noise with the magnitude of the error found to be a function of the magnitude of the frequency shift. Since the real covariance utilizes a signal heterodyned to a frequency centered on 125 Hz, frac- tional errors in the frequency estimate provide a larger error in the estimate of the differential frequency shift. 11.4.1.2 Complex covariance (1) The complex-covariance technique showed results unbiased by white noise for a sufficiently large number of samples. (2) The standard deviation was a function of the signal-to-noise ratio. The mean error was also within 1/VN of the standard deviation with N samples. (3) With pure noise, the frequency estimate was biased towards zero-shift because of the centering and non-flatness of the filters. (4) Errors in the frequency estimates were independent of the magnitude of the frequency shift from that of the carrier. 66 Doppler Shift From 1250 Hz 10 20 30 40 50 Figure 11.3. Real covariance error analy- sis showing error in Hz for given amount of Doppler shift from carrier frequency of 1250 Hz, as a function of signal-to- noise ratio (as defined in text.) I I I I I 1 £ 50 co 40 Frequency CO o J 20 CO u 75 10 c II Deviation o o 1 * Standard ro o - "O \^ ro -30 - No Signal Input b £ -40 c;n 1 1 1 1 1 ( ) 1 II 0.0 2.0 3 4 60 95 120 S/N Ratio 5.0 (rms volts) 14 (dB) Figure 11.4. Complex covariance error analysis showing errors and standard deviations of frequency estimates as a function of signal-to-noise ratio. 11.4.2 Consequences for Field Measurements (1) Under noisy conditions averaging times must be increased. (2) Wind variance measurements under windy (surface) conditions should be inspected carefully since the background noise level increases with surface wind speed. Thus, although the variance might be expected to increase with wind speed and the data might be observed to behave properly, such increases in the variance cannot neces- sarily be disassociated from S/N effects. However, for white noise, such variances in the frequency estimates will be equal to those obtained by using direct spectral techniques (R. J. Keeler, personal communication). 11.5 FIELD TESTS The results of the limited number of field tests to date have been encouraging. Following the construction of a prototype system during May 1979 we were able to obtain several nights of tower data with which to compare the new system's results. We found, at that time, zero bias in the mean values averaged over the total data set of three nights 67 and a mean difference in direction of 3 degrees. This system was then moved to a field site and has been running since 4 July with about 95 percent data recovery to a height of 600 m. A second device was then built several weeks before the BLIE experiment. During testing some biasing was evident. Further inspection showed two sources of error. The first was the presence of other transmitting frequencies. We found that sufficiently strong signals, although normally outside the bandwidth of our receiver, can pass through the wings of the filter and be aliased into the estimate of the mean spectrum. The choice of a second operating frequency for backscatter sounding must, therefore, be made care- fully. A second source of error was the presence of nonwhite noise produced by the tower. We analyzed the background noise with a real-time spectrum analyzer before, during, and after data runs. With a delayed trigger, spectra were averaged for signals characteristic of the 300-m range gate. Figure 11.5 shows a typical spectrum obtained with winds from the southwest greater than 4 m/s. Strong peaks in the noise are evident near 500 Hz and again near our operating frequency of 1250 Hz. The peak in this latter region occurred typically between 1100 and 1300 Hz, depending on conditions. We also calculated the errors in our measurements as a function of wind direction; the results are shown in Fig. 11.6. Underestimates are most prevalent with winds from the southwest. This is also the direction toward which the carriage support on the tower is oriented. This support con- sists of two grids of 1000 elements each spaced 0.3048 m apart with an opening of 0.267 m. At 25°C these dimensions correspond to acoustic frequencies of 1135 and 1297 Hz. 11.6 REFERENCES Sirmans, D., and B. Bumgarner (1975): Numerical comparison of five mean frequency estima- tors. J. Appl. Meteorol. 14:991-1003. Owens, E. J. (1977): Microcomputer-controlled acoustic echo sounder. NOAA Tech. Memo. ERL WPL-21, NOAA/ERL, Boulder, Colo., 76 pp. Neff, W. D., and E. H. Brown (1979): Acoustic echo sounder operations during PHOENIX, Chapter 14. In Project PHOENIX : The September 1978 Field Operation , W. H. Hooke (Ed.), NOAA/NCAR Boulder Atmospheric Observatory Rept. No. 1, available from NOAA/ERL, Boulder, Colo. 80303, and from NCAR Publications Office, Boulder, Colo. 80307. Kaimal, J. C, and D. A. Haugen (1977): An acoustic Doppler sounder for measuring wind profiles in the lower boundary layer. J. Appl. Meteorol. 16:1298-1305. 68 M Raw Echo /I / \ Bandpass Filtered Echo ^^ - ■ 1 2500 Hz Figure 11.5. Spectrum analysis of tower noise in the range to 2500 Hz with a 6 m/s wind from the south before and after band- pass filtering. 270< 180° Figure 11.6. Error analysis of winds (in m/s) showing the dependence of the errors on wind direction, which correlated with noise estimates made by spectral analysis. 69 12. DOPPLER ACOUSTIC SYSTEM FOR WIND PROFILING (AVIT) Paul MacCready AeroVironment Inc. Pasadena, California, U.S.A. 12. 1 INTRODUCTION The AVIT (AeroVironment Invisible Tower) system is a pulse Doppler acoustic unit continuously monitoring air motions aloft (mean winds and turbulence). The system para- meters have been chosen to provide all the atmospheric inputs for modeling the dispersion of atmospheric pollutants, on both research projects and operational programs. Considera- tions of economy, portability, simplicity of installation, reliability, and satisfactory operation in noisy environments are as important to the design as the basic accuracy and high-altitude capability. Since 1975, AeroVironment has developed and operated Doppler wind systems that use various antenna beam configurations, several Doppler shift analysis methods, and numerous transducer and enclosure designs. Before 1975 Ian Bourne at the University of Melbourne began development of high-altitude Doppler acoustic systems, with an emphasis on monostatic configurations and a full spectrum analysis technique for ascertaining Doppler shift. A formal collaboration between AeroVironment (AV) and the University of Melbourne began in 1978; AVIT is the system that evolved from this collaboration. The basic algo- rithms and electronics concepts used in AVIT were developed by Bourne. At AeroVironment, John Worden has been in charge of the AVIT program. Bourne's early two-component mono- static system is described by Bourne and Brann (1978), who also give examples of observa- tions and comparisons with radiosonde data. Hopper (1978) briefly reviews the system and also gives examples of measurements taken with it. 12.2. THE SYSTEM AND ITS PERFORMANCE AVIT is a flexible, modular system. The basic three-axis system uses three adjacent pencil-beam antennas. One tilts N (or S) 30° from the vertical to observe the N-S wind; one tilts E (or W) similarly to observe the E-W wind; and one points vertically to observe the vertical component. The antennas are operated sequentially, in the mono- static mode. Figure 12.1 shows a three-antenna array. Figure 12.2 shows a two-axis array mounted on a trailer. Figure 12.1. The three acoustic enclo- sures for the three-antenna system. Each contains a 1 . 8-m-diameter para- bolic reflector. 70 fflUMtilto Figure 12.2. Acoustic enclosures for a two-antenna system mounted on a trailer for easy portability. The en- closures are oriented vertically for transport. Parabolic reflector size is 1.2 m. A sound pulse (150 to 200 W) is transmitted at a frequency of 1500 Hz (2000 Hz also available), with a duration of 180 ms for the tilted beams. The received echo is heterodyned and then processed through an electronic comb filter with 31 teeth, to yield continuously the full spectrum. For each 33.3-m (100-ft) altitude range gate, the spec- trum is examined for acceptance or rejection, and then if accepted it is smoothed and curve-fitted, and the resulting peak frequency and amplitude are stored. At the end of the selected averaging period, say 20 min (variable from 5 to 30 min) , the assemblage of peaks is explored by a number of histograms, spectra, and interpolation techniques, and the best estimate of Doppler shift is ascertained along with an estimate of an observation reliability factor. The resulting wind profiles for each period are printed out on a Texas Instru- ments Silent 700 printer. In a three-axis system the display gives three components (vertical turbulence, and horizontal speed and direction) for every range gate, starting with 67 in (200 ft) and continuing up to the maximum height selected (up to 47 range gates) or the maximum height observed. The data can also be recorded on digital tape. Two other displays are available. One is a facsimile recorder giving a time plot of signal intensity vs. height. The signal intensity represents the strength of the echo after initial processing by the comb filter, i.e., it shows only "accepted" range gates. Thus this recorder conveniently displays the overall height and quality of the data . The other display is an oscillograph showing the spectra in real-time sequence for each range gate. It is a convenience when evaluating the contribution of ambient noise to the echoes. Finally, a speaker is available to make the echoes audible and to give the trained listener a good bit of information. Figure 12.3 shows a version of the complete data processing and display system. (Only item missing is the oscillograph.) The velocity component range is +15 m/s, limited by comb filter width. An automatic frequency alteration for the transmit pulse can be selected by the operator. Then, on the basis of the wind measurements during the preceding integration period the frequency is adjusted for the next period to keep the comb filter operating in its middle range. This extends the working range of the filter to component speeds of +25 m/s, and so in some directions wind speed can be covered to vector speeds of 35 m/s. This tech- nique permits the total frequency band evaluated to be narrow and minimizes spurious noise acceptance . Speeds are printed out to 0.1 m/s resolution, and directions to 3° resolution. Nominal reproducibility during periods of strong signals is deemed to be about 0.2 m/s, although examination of data runs often shows consistency between range gates to be within 0. 1 m/s and 3°. 71 Figure 12.3. AVIT data processing and display system in operation, with digital tape recording as well as data printout and facsimile display. The maximum altitudes at which data are obtained vary widely with meteorological conditions. An estimate from all operations with AVIT in Australia and the United States over the last year indicates that the 95% and 5% data recovery percentiles correspond to 300 m and 1000 m , with percent vs. altitude varying linearly between these heights. The clock has battery backup to maintain time accuracy during power outages. The system features automatic restart after power failure. It is also designed for unat- tended operation. 12.3. OUTPUTS Figure 12.4 presents a typical printout on the Texas Instruments Silent 700 printer. This is a 20-min interval display. The time during which data were taken is indicated at the top of the printout (date code, time 0622:02 to 0640:21). The left column gives height in units of 33.3 m (Range 2 = 66.7 m, Range 3 = 100 m, Range 9 = 300 m, etc.). Range 1 is omitted in this long-pulse, high-altitude version because there is insufficient time for the high-output driver to recover completely from its transmit power pulse before serving as the receiver transducer. Range is an option not included in this example; it gives the near-surface wind component information from two propeller anemometers mounted on a 10-m mast.) The numbers across the top give instrument operating codes and automatic gain control (AGC) levels. The columns from left to right are as follows: IIT NS R NO EW Height ranges in units of 33.3 m The N-S component (+ from N; m/s) Data reliability assessment for this component (0 is best, 9 is worst; data from to 7 are generally deemed suitable for meteorological application) . Number of pulses utilized in deriving N-S component. The E-W component (+ from E) . 72 STRRT TIME = ij ij 1 G| U6 \.\d END TIME= 1 Cl 06 4 : 1 as 1 1 34 ? ij 36 -, ■-. HT n: R ND EM R ND VERT R ND 5:1 VEL RNG R ND 33 -£. 7 3 7:4 ♦ -6. 1 ij 34 * 0. ij 1 34 Ij . z ♦ f, . t~\ 846 -:4 31 -8 . 3 3 31 * -5. 4 11 38 ♦ 0. 8 1 36 0. 3 ♦ £, . ij 84 3 _} 31 3 - 3 . 7 1 09 ♦ -4. 3 : : 28 ♦ ij. 8 8 3 0. 3 ♦ 5 7 8 8 3 7; 23 29 -4 . 2 -: _-: ♦• - 3 • 9 IJ 3 ♦ . 8 ~i cr 0. ] ♦ 5 . :-: 885 Ij 31 28 -4 cr 31 .: ; : : * — 3 . 6 1 88 ♦ -0. 1 30 0. 4 ♦■ tr . 3 819 £J 3 a? -4 .5 1 ;l J ♦ -8. 1 Ij 32 ♦ -0.1 8 27 0.4 ♦ c:: . Ij 8 04 1 32 26 -4 "7 Z" 3d ♦ -1 .4 31 ♦ . U 3 8 3 0. 3 ♦ 4 y 195 2 3 25 -5 "• cr " : j # - . 7 1 £8 ♦ 0. 1 31 0. 8 ♦ cr . 8 136 cr 3 24 -5 . 3 2 34 * -0.4 Ci J :'■ ♦ 0. 8 cr 38 0. 8 ♦ 5 •-■ 1 3 3 2 -; " c! _' — 6 . 2 ij : : : : ♦ 0. 3 34 ♦ 0. 1 _■ 34 0. 1 * 6 . 1 177 -; " 22 — 6 . 2 34 ♦■ . 2 : : _"' ♦ . ij 4 34 . 4 ♦ 6 . 2 1 3 ij "J ". 21 -6 . 1 ij 34 * 11. 3 II ♦ 0. 1 34 0. 1 •■ 6 , 1 177 ij -; - 2 -5 . 8 34 ♦■ 0. 2 Ij :4 •• . 8 32 0. 8 ♦ 6 > 1 3 ij ij 34 19 -5 3 ij 34 O - . 2 IJ 34 «► - ij . Cl 4 34 . 3 »• cr s 1 3 ij 34 18 -5 6 4 34 ♦ - . 3 1 34 ♦ 0. 3 4 3 4 0. 3 ♦ cr 6 136 4 34 17 -5 4 34 ♦ -1.9 Ij 34 * 0. 8 4 34 0. 8 ♦• c -? 1 98 ij 34 18 -5 3 3:4 ♦■ -8. 34 ♦ 0. 1 4 34 0. 3 ♦ cr J -7 801 34 15 -5. 3 34 # -1. 1 8 I-:4 ♦ 0. 4 1; ~; 0. 1 *> cr 4 198 t~_ 34 14 -5. 3 34 ♦ - . 2 11 34 ♦• - 1 j . 3 "7 "; "i 0. 8 ♦ cr -; 1 3 ij 34 13 -5. 6 8 ]:4 ♦• 0. 1 ij 34 ♦• 0. 1 cr ~; ~; 0. 1 •> cr J ■ 6 1 3 2 34 12 -6. 5 I-: 34 ♦ 0. 34 * . 4 3 8 0. 1 *■ t' . 4 1 3 "; 34 11 -6. 9 34 ♦ 0. 2 Cl 34 ♦ 0. 1 T. 34 0. 8 * 6 1 Q 1 3 Ij 34 10 -?. 3 34 *■ 0. 2 34 * . 1 t-. 34 0. 1 ♦ "7 2 1 3 Ij 34 9 -?. 4 1 34 * 0.4 IJ 34 ♦ 0. 4 34 0. ♦ 7 . ■-• 177 1 34 !-i — "^ 4 1 I-:4 ♦• . 5 3 34 ♦ 0. 1 4 34 0. 1 ♦ ■7 _ 4 177 q 34 ~7 _~7 8 2 34 *■ 0.4 u 34 ♦ 0. 1 l" 34 0. 1 ♦ cr 177 2 34 6 — 7" . 6 2 34 # . 5 ij 34 ♦ 0. 1 ~? 34 0. 1 ♦ 7 £1 177 lL 34 c -7 6 5 3:4 ♦ . 5 ij 34 ♦ . 1 cr -'. -'. 0. 1 ♦ "7 cr 177 5 34 4 -7. 3 1 34 ♦ . 8 z' 34 ♦ 0. 1 5 34 0. 1 ♦ "7 3 174 ■j! 34 3 - 6 > 9 34 ♦■ . 8 34 ♦ 0. 1 — 34 0. 1 ♦ ■7 ij 174 M 34 2 — 6 . 3 34 * 0. 5 L - «• - . 1 "■ 1 3 0. 4 ♦ 6 . 3 177 l! -1 ji Figure 12.4. Sample data printout of AVIT system. R Data reliability assessment for E-W component. NO Number of pulses utilized in deriving this component. VERT Vertical wind (+ denotes upcurrent) R Data reliability assessment for vertical component. NO Number of pulses utilized in deriving vertical wind. SD Standard deviation of vertical wind (calculated with respect to zero vertical wind). VEL Total horizontal wind speed (m/s). ANG Horizontal wind direction (degrees) R Data reliability assessment for speed and direction. NO Number of pulses utilized in deriving speed and direction. The operator commands a wide variety of printout options. Figure 12.4 repre- sents a common selection for operational uses, the main variations being different alti- tudes (the lower the altitudes, the faster the pulse repetition rate and so the greater 73 the amount of information available for processing); longer or shorter averaging times; and, for the two-component system, omitting the vertical component. In the example given, the horizontal data (components, speed, and direction) are derived from the horizontal components only. Since the vertical velocity averaged over 20 min is usually small, even in convective conditions, the error in omitting the vertical velocity correction is also usually small. The correction involves taking 1.73 times the vertical component, adding that to the N-S component and subtracting that from the E-W component (for the N-pointing and W-pointing antenna orientation employed in this example) to correct the components and hence the speed and direction. This correction can be selected in the computer and then the speed and direction printouts are for the corrected data. If the vertical velocity data observation is missing at that height, or has a bad reliability assessment number, the correction computation can assume a zero vertical velocity. For research purposes, the operator can choose to print out more information. The integrated full spectra for each component can be printed. Alternatively, the display will give the components calculated from each of six different histograms or spectra or combined analysis approaches. It is the agreement between these component estimates, plus a weighting depending on the number of pulses selected, that is used to derive the relia- bility assessment quantity. A combination of a specific histogram and spectrum were used to obtain the numbers presented in Fig. 12.4. The output can also be recorded on tape. We have used a Texas Instruments Silent 700 ASR cassette recorder for recording the results of each 20-min run, essentially the data illustrated in Fig. 12.4. We also use a Kennedy 1600 digital tape machine to record the integrated data and, if desired, to record fully every spectrum tooth output for every range gate for every pulse. This second option yields basic data for research on improving the algorithms for data selection and for exploring alternative methods of deriving turbulence from the echoes. 12.4. MAIN FEATURES The primary design options of a pulse Doppler acoustic system are (1) overall beam configuration, (2) transducer/antenna characteristics, and (3) the Doppler shift processing technique. (Features such as pulse generation, preamplifier, and data display, are not significant in differentiating systems.) For AVIT, the design choices are as f ol lows : (1) The overall beam configuration is monostatic. There is also a narrow angle bistatic variation available for special tasks. (2) The transducer/antenna system uses horn-driver-reflector-enclosure geometry and materials tailored from theory and considerable experimentation. In addition, a special treatment of the enclosure edges lessens diffraction at these edges, cuts sidelobes substantially, and makes the system suitable for use in noisy or urban locations . (3) Doppler shift for each pulse and range gate is ascertained in real time by a continuous full-spectrum technique coupled with versatile data selection/rejection criteria (and the data quality assessment factor is one of the system outputs). The following sections explore the rationale behind the first two design choices. The Doppler shift subject has already been treated briefly in previous sections. Suffice it to note here that the analog filter teeth (4.3 Hz wide) are both economical and stable; that the comb filter method substitutes for the more conventional full spectrum method, the digital FFT, while decreasing the demands on the computer; and that the comb filter method automatically weights the information in the previous range gate and adds it, to yield more significant information than that which can be derived solely by FFT examina- tion of echoes from a single range gate. 74 12.4.1 Overall Beam Configu ration In comparison with a bistatic beam configuration, the monostatic system offers several distinct advantages: (1) The antennas can be located adjacent to each other (even all on a single trailer) for convenience of installation. (2) The tilted monostatic antenna senses the same percentage of the horizontal wind at all heights. This is simpler than the bistatic case, which involves altered geometry at each range gate, and which has very strong sensitivity to vertical com- ponents at the high range gates. (3) Only pencil-beam antennas are used in the typical monostatic system, while fan beams are used in bistatic systems. The broad fan beams have poorer sidelobe suppres- sion, even when used with complex enclosures. When used for receiving, the fan beams are relatively inefficient in keeping out noise from low elevation angles; when used for transmitting, they can be disturbing to persons nearby. (4) Tilted beams can be tilted away from noise sources, to minimize interference. There is one significant disadvantage to the monostatic system. It operates on echoes scattered only at 180°, and such scattering comes only from the temperature micro- structure field. The bistatic systems utilize echoes from both the velocity and tempera- ture fields. With moderate turbulence but near-neutral stability, one would expect mono- static scattering to be weak whereas bistatic scattering would be strong. Another disad- vantage, but one which is generally insignificant, is that the sensed volumes at a given height are not at exactly the same location. This is of little concern when data are averaged over a few minutes, since the data are then deemed representations of an air volume large in horizontal extent. It turns out that in virtually all conditions the monostatic approach is practi- cal because the high efficiency inherent in having the transmit and receive beams line up exactly compensates partially for the weaker scattering cross section. Thus the main disadvantage of the monostatic system is not overwhelming; one can use the system and benefit from its good features. The average altitudes reached (see Section 12.2) demon- strate the general suitability of the technique. At night, with stability aloft, partic- ularly in complex terrain, there are good monostatic signals at 500 m even with very light turbulence (less than e 1 ' 3 = 0.5 cm 2 ' 3 s _1 where £ is the dissipation rate). The worst period for monostatic echoes (and bistatic) is typically in weak pressure gradient condi- tions in late afternoon when solar heating shuts off, heat flux up into the atmosphere stops, the lapse rate remains close to neutral, and the turbulence dissipates. To cover this period we have tested an approach that merges the bistatic method witli a monostatic system . 12.4.2 Antenna Systems AVIT employs a standard driver feeding downward into a tuned horn, spreading sound out to a parabolic reflector from which a pencil beam is emitted upward. The assem- bly is housed in a large acoustic enclosure that decreases the sidelobes for transmitting or receiving. The research instrument configuration uses a 1.8-m reflector, and a parallel- sided enclosure built of six sheets of 1.2-m by 2 . 4-m plywood or chip board. The enclo- sure is lined with 5-cm-thick glass wool for sound absorption. The glass wool is used instead of plastic foam because rain water drains better with the wool. The parabolic reflector can be heated for snow removal. The two-component portable instrument version uses 1.2-m reflectors in five- sided enclosures, with both enclosures optionally mounted on a trailer for easy portabil- ity. During transit the enclosures are set upright. They are then set up for operation by tilting 30° from vertical in the N and W directions (or other orthogonal orientation). 75 The top edges of the enclosure are equipped with Thanadners, at Lockheed Co. Thanadners are teeth, here 0.6 m high and 0.1 m wide, covered on both sides with absorbing material, which have the effect of acoustically "feathering" the top edges and greatly decreasing diffraction. Our tests have shown the additional sidelobe suppressions at low angles to be as much as 12-14 dB. The amount will depend greatly on the sidelobes ini- tially present, and on whether or not the wall attenuation is adequate. The Thanadners help make the acoustic system suitable for operation in urban environments. The wind speed at which wind noise becomes a problem has not been determined for these enclosures, either with or without Thanadners. On several installations we worried about possible wind noise, but avoided the problem by erecting a fence upwind. 12.5. THE FUTURE AVIT is satisfactory as a high-altitude remote probe for many purposes, but improvements are always desirable. Work will continue in order to reach higher altitudes or to cover a given altitude a larger percentage of the time. The main approaches are to increase power and to apply more sophisticate*d algorithms for extracting valid Doppler shift information when the signal/noise ratio is low. Work will also continue to obtain additional turbulence parameters, and to validate the observations by aircraft measure- ments . Our goal is to have an instrument that provides all the meteorological inputs needed to supply a rational diffusion model covering a wide range of conditions. Wind and turbulence profiles are obviously essential inputs. Temperature gradients are needed only to assist in calculating plume rise, and they need not be precise; the standard plume rise equation can use only a few broad categories, such as unstable, neutral, stable, and very stable. It may prove possible in many cases to derive such categories from the profiles of wind and turbulence. 12.6. REFERENCES Bourne, I. A., and H. N. Brann (1978): AIRMET Conf. R. Met. Soc. (Aust.), Bureau of Meteorology, Melbourne, Australia. Hopper, V. D. (1978): Acoustic sounding of the atmosphere. Endeavour (new series) 2:121. 76 13. RADIAN CORPORATION MODEL 800 ECHOSONDE M. A. McAnally Radian Corporation Austin, Texas, U.S.A. 13. 1 INTRODUCTION The Doppler acoustic sounding system is used to characterize the thermal struc- ture and wind profiles in the lower atmosphere (below 1 km). Transmitted acoustic tones are scattered by the turbulent atmosphere. The strength of the direct backscatter depends on the temperature fluctuations in the scattering volume. The frequency of the scattered energy is shifted by an amount dependent on the motion of the scattering volume. There- fore the returned echo strength as a function of time provides information about the thermal structure as a function of altitude. The mean Doppler frequency shift is propor- tional to the average velocity of the scattering volume along a line that bisects the transmit and receive beams. By measuring the mean Doppler frequency shift in three inde- pendent directions, the three wind components can be determined as a function of altitude. Only recently with the availability of low-cost digital systems has it been possible to estimate the Doppler frequency shift in a cost-effective system. Radian's Model 800 Doppler Echosonde utilizes the LSI-11 microcomputer to perform complex covariant process- ing to measure the Doppler frequency shift of the returned echo. The Radian Echosonde system comprises five basic subsystems: (1) Acoustic antenna and transducer assembly. (2) Acoustic noise suppression (Septacuff). (3) Bistatic transmit horn. (4) Microcomputer and control electronics. (5) Display terminal. 13.2 ACOUSTIC ANTENNA ASSEMBLY The antenna assembly consists of an exponential horn and transducer which di- rects acoustic energy into a parabolic reflector to form a 10° acoustic beam. Because of the high magnetic field strength in the compression driver coil gap, the assembly also functions well as a return echo detector. The electrical signal from the compression driver-detector is amplified at the antenna to minimize electrical noise effects on the receiver system. The transmitted tone is 2000 Hz in the standard configuration; however, other frequencies have been used. The transmitted tone power is 150-W electric input power with a selectable pulse duration of to 990 ins in 10-ms steps. The pulse repetition rate is also selectable from 1 to 99 s in 1-s steps. The pulse duration is the controlling vari- able for altitude resolution, and the repetition rate sets the maximum travel time, and thus the maximum altitude. 13.3 ACOUSTIC NOISE SUPPRESSION The acoustic enclosure design, based on theoretical and experimental studies by the NOAA Wave Propagation Laboratory, isolates the receiving antenna from ground-based 7 7 interfering noise sources. The fully portable Septacuff is constructed of seven sides and a bottom, all of which are lined with acoustic foam and a sandwiched lead sheet septum for maximum sound control. The Septacuff shape was selected to minimize diffracted ground- level noise. The sides of the enclosure are flared away from the 1.2-m parabolic antenna/ reflector at the bottom. The total weight of the enclosure is 250 kg (550 lb); it can be disassembled down to parts that weigh less than 50 kg (110 lb). 13.4 BISTATIC TRANSMIT HORN The bistatic exponential transmit horn is constructed of fiberglass and shaped to form a broad vertical beam to provide information over a range of altitudes. The bistatic transmitter uses the same type of transducer as the monostatic antenna assembly. The transmitter provides a fan-shaped beam with 50° vertical beamwidth, and 10° horizontal width . 13.5 MICROCOMPUTER AND CONTROL ELECTRONICS This subsystem consists of a microcomputer, amplifiers and filters, controls, and power supply. The central processing unit is a Digital Equipment Corporation LSI-11 with 4-K memory and hardware multiply/divide. Because the control program resides in programmable read-only memory (PROM), the system will automatically restart after power failure. The Echosonde signal-generating and processing electronics are assembled on printed circuit boards designed for plug compatibility with the LSI-11 microcomputer. The modular system design provides many optional system configurations by exchanging or adding circuit boards. For example, a user may select the option that provides only single axis measurement, i.e., only the vertical wind components and temperature structure, and up- grade to a full three-axis system by adding the bistatic transmitters, and changing the PROM board, which contains the system software. 13.6 DISPLAY TERMINAL The display unit is a dot-matrix digitally controlled line printer. The printer has a special optical shading character set for displaying the intensity of the back- scatter return. Figure 13.1 shows a sample of the display terminal output. The standard alpha-numeric character set is used to print the wind data determined by the Doppler frequency shift. The figure shows horizontal wind speed (tenths of m/s), horizontal direction (degrees east of north), vertical and horizontal wind direction variance (de- grees), and a vertical wind indicator. The display can also provide an estimate of the returned signal -to-noise ratio. 13.7 OPTIONS In addition to the basic subsystems, the Echosonde system offers the following optional features: (1) 220 V a.c, 50-Hz input power. (2) Telecommunications (serial ASCII) over two-wire or standard telephone with standard modems. (3) Optional peripheral storage units including 9-track magnetic tape, cassette tape, disk, or paper tape. (4) Heated antenna/ reflector . (5) Tilting base for Septacuff. 78 5SJH — BKgpnsir /'It. 4 : 50 iREEd v DIR 75 TLMF- 222 Fl UX d ■ r- ' *$»? < ill Si™ ... * TjfJv "":• Tiiy 22 i'2 3» 27 i2v'» 2T"i2i:i« 21 i;v« iv n** 1-0 i32« is iis« it: 1 ss»~'2 i'*T^^frfJBo» 21 147* si mi. 22 ?«*"¥■■»« I ?'S:"'7 "".;. i" "i '■ S » 2*'-'' < 1 t IS 1 fr 2 IV. i s "2 ,0 """»:O5 2""'.„. v 2 6 12" 7/16 5: o SPEED 12 CUR frfc rtl-U 222 FLUX Sr7«i«VK"^ : - v "2- ; "f'# '" : f;":fc"^ iv^ -#- ?«■-.•• i 4 ,, 4 .. ii;,:-'S*'., ft ".Vr.>j: i ■.;;■;::;■ ■$ 'S'-javj " -B -o ■.•;-«,_ -)"• y ™'.l* ^ r !.2i* 52 .. ji.*« ?.(_ 12'« 'h l2v» 2t !*•*• 25 |"5> 25 2/. ;■/ I '£« 22 12** 24 I4I» 22 1 29« 6 210* 17 ; i'C5f. ,! 3lSC.""4F> ?M1 t r fi'^i" ^ : ' : "^ m ft' [n- cwitf t;: 4';«3 ^S iJ : :■ i < i 7 i 1 1 j ;;*-: >r 1 da.A«*j-4-.. -E ■!■ —■ ■>:-[■*:.:'■„ ft C ft :. A' ::: C ::•-*=«. A 7/ifc 5: U iPEED 8 DIR 57 TEMR 22c Fl MX 11 4!"'2 '-■ 5- j..-"<"?.S ! ;;.:.:: h ;• 1^2- 2n 5 "d" ' A -D s Y-ii^: M, :>'.■ ft A --•«-• E - A ' . U. is 2J E . C " • B ;■:'■■ 7»«i»*^«~.W-.7S*' :; - £ , ^ iS** -t I-'* "•it?/' 1i"\Tf'» 2t I2v« Iv'"|il7« 21 144« >3 1 4 1 » 21 156« 24 14l» 6 210» 7 264« 20 265« 11 2 S: 1S»*I 2*- V*»3T2-" iV-y.. fr 1 .•"».' 7„. 1 11 2. . % ,ft*.fc!f: s* B**ff:,|.. fl H n r. 9 "i""Ti"5~'i 2ii 3 "TK .'"■■ I* 1 v ', 17»: i... 27- : S- -'■ 1 ""20**' - - '■v*"w~ : S m . V Figure 13.1. Sample format of display from printer. 13.8 SYSTEM CONFIGURATION OPTIONS The optimum acoustic sounder configuration is dependent on the specific require- ments of the application. The user of the acoustic sounder system must consider the following : (1) Altitude range required. (2) Sampling interval required. (3) Noise interference as well as potential noise nuisance. Radian's Model 800 Doppler Echosonde can be configured in a variety of ways tailored to each specific application. All configurations determine thermal structure and display on the line printer, but the user may elect to record the following: (1) Single-axis (vertical) wind component. (2) Two-axis (vertical and one horizontal) wind components. !<) Plan view Monostatic transmit receive Monostatic transmit receive ^~^ Plan view Monostatic transmit receive Figure 13.2. Optional three-axis (u, v, w) configurations for the Model 800 Doppler Echosonde: a) bistatic, b) monostatic. (3) Three-axis (vertical plus orthogonal horizontal) wind components. Figure 13.2 illustrates two choices for arranging the transmit/ receive antennas to record wind components. The bistatic arrangement separates the transmit and receive antennas for resolving the horizontal wind components. This arrangement provides the simplicity of receiving all returned echoes at one receiver. The monostatic arrangement provides one to three transmit/receive modules complete with acoustic enclosure. Two of the acoustic enclosures are tilted to obtain the horizontal wind components. The monosta- tic arrangement has the advantage of directional focusing of the transmitted acoustic energy . 13.9 CALIBRATION VERIFICATION The Model 800 Echosonde normally requires no field calibration after installa- tion. The instrument is fully calibrated and tested prior to shipment. It is usually advantageous, however, to verify several critical aspects of the Echosonde operation following a field installation. The sound level of each of the acoustic transmitters is normally checked after installation. The sound level of each bistatic transmitter should be 133 dB relative to 20 (jN/m 2 . The monostatic transmitter sound level should be 128-130 dB relative to 20 (JN/m 2 . The preamplifier d.c. offset is also normally checked after installation. This is accomplished by injecting a 100-fjV signal into the preamplifier and adjusting the zero trim pots for less than 1-mV offset voltage. The remaining analog adjustments are not usually possible in the field without the use of specialized equipment, 80 14. SWEDISH SODAR SYSTEM Soren Salomonsson Department of Meteorology University of Uppsala Uppsala, Sweden Mats Hurtig Sensitron AB Stockholm, Sweden 14. ] INTRODUCTION During the period 1970-1975 a monostatic acoustic sounding system was developed at the Swedish Research Institute of National Defence under the leadership of Hans Ottersten This sodar system was then utilized in research projects in Sweden and other European countries (Ottersten and Eklund, 1973; Ottersten et al., 1974; Ottersten, 1975a and 1975b). Starting in 1975, the Swedish sodar system was further developed in joint projects between the Department of Meteorology of the University of Uppsala and the electronic company Sensitron AB of Stockholm (Holmgren et al., 1976). A new monostatic Doppler system was built in 1976 by applying a phase-locked loop (PLL) circuit for tracking the frequency shift of the return signal. The sodar project is supported by the Swedish Board for Space Activities and the Swedish Board of Technical Development. A first application of the Doppler unit was done in cooperation with the Swedish Meteorological and Hydrological Institute in a boundary layer study during the first two weeks of May 1977. The vertical wind velocity was derived from the Doppler shift at 100, 175, 250, and 350 m at some brief periods. Figure 14.1 shows a set of these measurements. m/s 0.4 2 -0.2 -0.4 4 0.2 -0 2 -0.4 4 0.2 -0.2 -0.4 0.4 0.2 -0.2 -0.4 : / -■—a' _i i \i 1 \/^V ' \/ Xv -x — X— X— ,__ / ^* J ! I I I I I 1 I I I l_ \ I \ I / I I I I I _j 1 1 *_- \ -"^i y --\ / LEVEL 350 m LEVEL 250 m Jt^ LEVEL -S ^-X. — y 175m ■^ 'Nu/' LEVEL -x X 100 m _!_ _L _l_ _1_ J_ _L _I_ _u _L LOCAL TIME 10.56.40 56.50 5700 57.10 57.20 57.30 57.40 57.50 58.00 Figure 14.1. Vertical wind velocities derived from the acoustic Doppler measurements for different heights on 5 May 1977. 8 1 WIND VELOCITY (m/s) 180 _L LOCAL TIME _i_ 12.21 25 29 33 37 41 45 49 53 57 13 01 05 09 13 17 21 25 29 33 37 41 45 49 53 571401 Figure 14.2. Example of the horizontal wind velocity and wind directions measured by Doppler (solid line) and pibal tracking (crosses) at a height of 60 m. Each data point is centered on the corresponding 2-min averaging interval . A preliminary description of the sodar measurements obtained during this field project is given in Salomonsson and Ivarsson (1978). Measurements of the horizontal wind velocity were carried out during the autumn of 1978 for the benefit of a project that aimed at finding the best location for a wind power station in the area. Two monostatic sodar systems, with tilted antennas, were used to determine the horizontal wind components. The derived wind velocity was compared with simultaneous wind measurements obtained from double theodolite pibal trackings. An example of these measurements is shown in Fig. 14.2. 14.2 TECHNICAL DESCRIPTION The monostatic system used at the intercompa ri son in Boulder is a commercial system manufactured by Sensitron AB. It consists of two antennas for measuring the two horizontal wind components. Figure 14.3 shows the relationships of the parts of the system. Table 14.1 lists the specifications. The system uses two antennas, but a third antenna can be added to measure the vertical wind velocity. The antennas can be clustered together, with the tilting antennas pointing out from the central point along the orthogonal planes, or they can be separated as shown in Fig. 14.4. The antennas are fed by compressor drivers of 100 W. They give an acoustic beam pattern like that in Fig. 14.5. 14.2.1 Description of the Monostatic System During transmission, a tone-burst is generated in the transceiver unit SR 20 by the combined band-pass filter and tone generator. The tone-burst is fed through a power amplifier to the transducer of one of the selected antenna channels. The received echo signals are amplified in the preamplifier and transferred by balanced cables to the receiver in the transceiver unit SR 20. 82 Antenna Units SR 40 #%A Transceiver UnitSR 20 SwitchboxSR80 Signal/ Noise Analyze! T/R Switch Pre- amplifier $ > i> High Pass Filter AC Amplifier HH Doppler Unit PLL 1/R Amplifier Power Amplifier BP Filter and Tone Gen. Ramp Interface SR90 f Digital Tape Recorder SR 36 — $-$-$— - Tape Driver Detector Control Unit I I Log amplifier Computer Chart Recorder SR32 Color Display SR34 O o I Keyboard Figure 14.3. Block diagram of the acoustic Doppler sounder. The echo signals are high-pass filtered to eliminate high energy components of the background noise at frequencies below 800 Hz. A linear-gain amplifier is used to compensate for spherical divergence of the scattered acoustic wave. The PLL circuit gives an output signal proportional to the Doppler shift of the incoming signal. To reduce noise the echo signals are passed through a signal - to-noise analyzer, which selects signals that have a given ratio to the background noise level. The detected Doppler signal is connected by the interface SR 90 to the microprocessor in the tape recorder unit SR 36 for determination of the wind vector. The other channel works in exactly the same way. The wind information received and processed by the microprocessor is presented on the color display SR 34 (Thomson and Scheib, 1978). The intensity for the temperature fluctuation received at one of the antennas is band-pass filtered and detected. The signals are processed and presented on the color display SR 34. Even chart recorder SR 32 presents temperature fluctuation and is fed by a log amplifier, which compensates for the difference in the dynamic range of the detected signal to the recorder. 14.2.2 Data Presentation The color display SR 34 is an eight-color CRT screen that contains a micro- processor for the color representation. The wind information is presented on the screen in different pictures which can be selected from the keyboard. Profiles of wind speed and direction, columns of wind speed and direction, and a combination of wind speed and the intensity of the temperature fluctuation can be displayed. The intensity of the tempera- ture fluctuations is classified in five colors and presented in a height/time diagram. Integration time of data, color representation, program start-up, etc., can be controlled from the keyboard. 83 Table 14.1. Technical specifications of the Swedish sodar system Equipment Specifications Antennas (two) Parabolic dish Fiberglass Diameter : 1.2m Transducer Altec Lansing 291-16B Transceiver SR 20 Transmitter Frequency: 2,400 Hz (1,800 Hz) Pulse power: 100 W Pulse width: 30, 90, 180 ms Pulse repetition rate: 0.1 - 0.2 Hz Receiver Balanced input: 600 Q High pass filter: 800 Hz Range correction: 1/R within height range 20-1,000 m Bandwidth: 20, 40, 80 Hz (for recorder SR 32) Sounding range: 20-1,500 m PRE AMP PRE AMP TRANSCEIVER A TRANSCEIVER 8 THREE CHANNEL RECORDER WIND PP0CES SING UNIT MtrIG DELAY Ij flwlN TRIG DELAVlJ I YJWIND PR0CES- UNIT " SING UNIT THREE CHANNEL RECORDER SODAR RECORDER SODAR A SODAR B Figure 14.4. Block diagram of the monostatic Doppler system used in the pro- ject on Gotland and the horizontal orientation of the two antennas (A and B) in relation to the north axis. 84 Figure 14.5. Polar pattern of the acoustic beam for the transmitting antenna at 2,400 Hz. (The lobe width is about 8°.) All data are recorded on tape in data blocks. Every block contains number of day, time, wind information, and intensity of the temperature fluctuations for one sounding interval . Another presentation unit is recorder SR 32 which gives the intensity of the temperature fluctuation along one axis. Recorder SR 32 works on a new principle, present- ing grey shades on metallized paper. This recorder is almost free of maintenance compared with earlier recorders. 14.3 EXAMPLES OF DOPPLER MEASUREMENTS Figure 14.1 shows graphs of vertical wind velocities derived from the Doppler shift at four selected levels obtained in an atmosphere of a slightly stable thermal strat- ification. It may be noted that the measuring period is only about 1.5 min. Therefore, a discussion of average wind velocities has no significance in this case. During ideal conditions one may expect to get zero mean vertical velocity if the averaging interval is long enough (Kaimal and Haugen, 1977). Variations in the vertical velocities indicate a "wavy" pattern with a tendency for crests and troughs to coincide at the four levels. In order to discuss the vertical wind structure in detail a much longer tune series is needed. During the autumn of 1978 an extensive project aimed at finding the best location for a wind power station was carried out in selected high-wind areas on the Swedish island Gotland (Smedman-Hbgs t rom and Faxen, 1980). In connection with this wind-prospecting project, supported by the Swedish National Board for Energy Source Development, two mono- static sodar systems were used. The sodar measurements were carried out in a joint project between the Department of Meteorology of the University of Uppsala and the Swedish Meteor- ological and Hydrological Institute, with technical support by Sensitron AB . A simplified block diagram of the two monostatic sodar systems (A and B) used in the wind-prospecting project is shown in Fig. 14.4. The system was built around two stand- ard sodar units. The Doppler frequency received at each antenna was analyzed in the wind- processing unit and presented on the analog 3-channel recorder. The two sodar systems operated at the same frequency (2.4 kHz). To avoid interference between the two systems each transmitter was pulsed separately in time by the trigger delay unit, with a pulse repetition frequency of 0.25 s . The antennas were tilted 50 degrees from the horizontal plane, with the intersection of the antenna beams at a height of 60 m . The radial velocities were derived from the 3-channel recorders. By using the assumption of a zero mean vertical velocity for the time-averaging interval and trigonomet- ric relationships the horizontal wind vector was computed. Figure 14.2 shows a preliminary S r , result of the horizontal wind measurements. The wind velocity derived from the Doppler shift was then compared with simultaneous wind measurements from double theodolite pibal trackings (Alexandersson and Bergstrom, 1979). Wind velocities show satisfactory agreement, but the wind directions show more differences. These differences could probably be explained to some extent by considering the uncertainties in the determination of the Doppler shift (Beran and Clifford, 1972). A small error in the determination of the Doppler frequency in one or both of the components affects the wind direction more than the wind velocity. Another reason for the discrepancy is that in general the actual position of the balloon does not coincide with the intersec- tion of the antenna beams. Furthermore, the vertical wind velocity will also influence the measurements, especially when the averaging period is not long enough to make the mean vertical component negligible. A more detailed analysis of the whole series of tests of the Doppler measurements from the wind-prospecting project will be published in the series of reports from the Department of Meteorology of the University of Uppsala. 14.4 REFERENCES Alexandersson, H., and H. Bergstrom (1979): Evaluation of double theodolite pibal track- ing data. Report No. 55, Dept. of Meteorol., Univ. of Uppsala, Uppsala, Sweden. Beran, D. W. , and Clifford, S. F. (1972): Acoustic Doppler measurements of the total wind vector. Preprints AMS 2nd Symp . on Meteorol. Obs . and Instrum., San Diego, Calif., 27-30 March 1972. American Meteorological Society, Boston, Mass., pp. 412-417. Holmgren, B., C. Jacobsson, and H. Ottersten (1976): Sondering av atmosfarens gransskikt med vertikalsodar . FOA rapport C 30077-E1, National Defence Research Institute, Stockholm, Sweden. Kaimal, J. C., and D. A. Haugen (1977): An acoustic Doppler sounder for measuring wind profiles in the lower boundary layer. J. Appl. Meteorol. 16:1298-1305. Ottersten, H. (1975a): Fjarranalys av atmosfarens gransskikt med sodar och radar. Styrel- sen for Teknisk Utveckling. Slutrapport STU 71-727/U 957b, April 1975. Ottersten, H. (1975b): Swedish sodar investigations and results. Paper presented at the URSI XVIIIth General Assembly, Lima, Peru, August 1975, International Union of Radio Science, Brussels, Belgium. Ottersten, H., and F. Eklund (1973): Remote sensing av troposfaren. Styrelsen for Teknisk Utveckling. Lagesrapport STU 71-727/U 597. Ottersten, H., M. Hurtig, G. Stilke, B. Brummer, and G. Peters (1974): Shipborne sodar measurements during JONSWAP II. J. Geophys . Res. 79:5573-5584. Salomonsson, S., and J. Ivarsson (1978): Sodar measurements of the boundary layer during the field project "Stenungsund-77 . " Report No. 50, Dept. of Meteorol., Univ. of Uppsala, Uppsala, Sweden. Smedman-Hb'gstrb'm , A.-S., and T. Faxen (1980): To be published in report series. Dept. of Meteorol., Univ. of Uppsala, Uppsala, Sweden. Thomson , D. W., and J. P. Scheib (1978): Improved display techniques for sodar measure- ments. Bui!. Am. Meteorol. Soc. 59:147-152. 15. THE XONDAR Robert L. Peace, Jr. Xonics, Inc. Van Nuys , California, U.S.A. 15. 1 INTRODUCTION The XONDAR (Xonics Doppler acoustic radar) is an operational, commercially avail- able system. It is designed for a broad range of applications, from mean wind measurements to turbulence and diffusion studies. It can also be used in environments ranging from benign to hostile. The XONDAR can vary several key parameters, such as pulse length, pulse repetition frequency, summation time, and observing altitudes, in the field with a simple computer command. It can also suppress the effects of ambient noise (Balser et al., 1976a) Optional equipment includes a sounder printer (to display the time-height distribution of relative stability), hardware to melt ice and snow, and telephone or radio modems. 15.2 THE BASIC XONDAR ACOUSTIC ANEMOMETER A basic XONDAR wind-profile measuring system consists of five subsystems: the sound-producing and -sensing hardware, the computer, the printer, the program input and storage device, and the command keyboard. Two models of the XONDAR are offered, the Model 300 and the Model 600. These differ primarily in hardware attributes that give the first system finer vertical, temporal, and velocity resolution, and the second system a greater altitude range. The characteristics and attributes of both models are given in Table 15.1, although only the longer range Model 600 participated in the intercomparison . Table 15.1. XONDAR wind sensor technical specifications Characteristic Model 300 Model 600 Antenna diameter (inside) 3 ft (0.9 m) 4 ft (1.2 m) Peak power to transducer (electrical) 250 W 250 W Peak output power (acoustic) 40 W 80 W Transmitted frequency 4000 Hz 2000 Hz Pulse duration (computer-controlled) 80/40 m s 160/80 m s Altitude resolution 25/12 m 50/25 m Velocity accuracy (one component) 0.2/0.4 m/s 0.2/0.4 m/s Maximum velocity (one component) +25 m/s +25 m/s Pulse repetition interval (normal) 2.5 s 5s Nominal maximum altitude for velocity 200 m 500 m Maximum altitude for sounder 400 m 800 m Number of altitudes sampled 10 10 Averaging time (typical) 2 min 2 min Power requirements 115 V +10% @ 30 A, 50-60 Hz PROCESSOR SITE ANTENNA SITES Crystal Oscillator . 4- Timing Signals Pulse Gate -> Line Driver Power Amplifier — ► T/R SW. <-> Transmit/ Receive Antenna ^_From Comjputerj 1 Preamplifier #3 4— 1 | ■ ' " ' J Analog/Digital Converter 4- Multiplexer <- Signal Conditioner *-» Preamplifier #1 Receiver Antenna 1 *n • 4 1 1 r Output Signal 1 1 Preamplifier #2 *- Receiver Antenna 2 [_To Computer J Figure 15.1. Block diagram of XONDAR sensor subsystem. Figure 15.2. (Above) The three XONDAR antenna housings. (Right) The stand- ard antenna configuration of a Xonics three-component air-motion measuring system. 1 KANSXim ER/REC'F.IVER ) RECEIVER I 15.2. 1 The Sensor Subsystem The sensor subsystem for both models of XONDAR is composed of one transceiving and two receiving antennas and their supporting electronics (Fig. 15.1). Each of the three antenna assemblies consists of an aluminum parabolic reflector mounted in the bottom of a foam rubber and lead-lined acoustic shield (Fig. 15.2, above). This shield effectively prevents sound from escaping from the transceiver housing or entering any of the housings 1 Parameter Inputs fpulse Signal L Timing (Including Real-Time Clock) Output ' • To Transmitter 1 Sajnpling Signals ' > Device 1 . . i ' 1 K ' (_To Receivers •t » i r i 1 Digitized i Buffer Storage FFT Memo iy Decision Logic ! Received Signals ' |_ | Figure 15.3. Block diagram of basic computer functions except in the desired direction. At the focal point of the transceiving antenna is an acoustic transducer that both emits high-power pulses of sound in a narrow circular beam and detects sound returning from the atmosphere. The two outlying receiver antennas are equipped with transducers and feedhorns that detect sound from a range of angles that is narrow in the horizontal but fan-shaped in the vertical and centered on the transmitter beam boresight. In operation, the three XONDAR antennas are nominally located in the right- triangle configuration diagrammed in Fig. 15.2 (right) with an antenna separation equal to, or somewhat less than, the maximum altitude to be observed. To avoid loss of sensitivity of the horizontal components of air motion, the elevation of the receiving antennas should not exceed 50° to 55°. For the nominal 500-m maximum altitude of the Model 600, antenna spacing should be 350 to 400 m. Although the antenna specifications and the configuration shown in Fig. 15.2 (right) are optimum, the computer program that performs data reduction is sufficiently versatile to allow considerable variation in antenna spacing, orientation, and relative height where site constraints dictate. The electrical power requirements of the XONDAR systems are 100-120 V, 50-60 cycles at 30 A, supplied only at the computer location. The low-voltage power requirements of the receiving antennas are provided through the interconnecting cables supplied with the system. These cables also carry all intelligence to the computer and commands to the antennas . 15.2.2 The Computer Subsystem The heart and brain of the XONDAR system are a minicomputer with 32,000 words of memory, a real-time calendar clock, a precision lapsed-time clock, and a nonvolatile memory sufficient to read in and restart the program after a power failure. Figure 15.3 is a diagram of the basic functions performed by the computer. The system has considerable flexibility in its configuration, operational mode, and data format. The standard selectable system configuration elements are relative antenna height, spacing, and orientation. Selectable operational modes are pulse duration (which determines velocity and height resolution), pulse repetition frequency (which determines maximum unambiguous observable height as well as the number of samples per summation period), integration interval (which must balance the frequency of observations against the statistical significance of the data), duration of observation versus no observation (if intermittent operation is desirable), and height levels to be sampled (up to 10). The data 89 output for the basic system can be in either the vector component form (V , V , V„. o\,, a,,, . r x x V 2' X' y' a z ), with or without the standard deviation of the velocity component (o x , o , o z ), or the more common polar coordinate form (A z , S p , E-^, o a , a s , a^) , with or without "the standard deviations. Additional types of output are available, but they were not used for this intercomparison . 15.2.3 The Program-Storage Subsystem The computer program to control the XONDAR's operations, including the default values of the system configuration, operational mode, and output data format, is provided on a floppy disk. This disk is read into the computer each time the system is activated after a period of disuse, or when changes in the program or default parameters are desired. Once read in, the program automatically begins operation of the XONDAR . if power is inter- rupted during operation, the disk drive automatically rereads the program and default parameters into the computer and begins operation whenever power resumes. The real-time calendar clock continues to operate for long periods on its own storage battery. Thus, the system resumes operation without loss of date or time reference. The residence of both the controlling program and a default value of all opera- tional parameters on a floppy disk makes it possible for Xonics to rapidly and economically provide a user with different default values whenever the basic XONDAR use changes. If a variety of operational modes or system configurations is anticipated, Xonics can provide a diskette for each. These are easily inserted into the reader. Thereafter the system will operate in accordance with the new instructions until a different diskette is inserted or parameters are changed through the keyboard. Should it be desirable to add optional hard- ware or data-analysis capabilities, the computer program necessary to accomplish the changes is also provided by Xonics on a system-compatible floppy disk. 15.2.4 The Printer The system printer is a stand-alone 40-column printer connected to the computer through a simple cable. This arrangement makes it possible to locate the printer wherever most convenient within several feet of the computer. The unit prints data, heading informa- tion, and command prompts and echoes keyboard entries on a 3^-in-wide roll of ordinary paper. A take-up reel rewinds the paper a few inches beyond the print roller. Sufficient paper is exposed to allow examination of two complete 10-level sets of air-motion data without removing the take-up reel from the printer. 15.2.5 The Keyboard All operational communication with the XONDAR system is accomplished through a standard computer-type keyboard mounted in the computer cabinet. Temporary changes in one or more of the operating commands or parameters are easily entered through the keyboard. The new command or parameter overrides the corresponding value in the computer. The new value is then used until it is changed through the keyboard or overridden by the contents of the diskette subsequent to a power interruption or an operator command to read the diskette . 15.3 PREVIOUS TESTS OF THE XONDAR This intercomparison of acoustic Doppler radar systems is not the first for Xonics systems. The earliest prototype XONDAR (a three-antenna, bistatic, two-component system) was compared with cup anemometers and wind vanes mounted on a 150-m tower at White Sands Missile Range in 1973 (Balser et al., 1976b). In October 1974, a four-antenna, fully bistatic system (made for the Air Force Cambridge Research Laboratory) was compared with anemometers on a 500-m tower at the Atomic Energy Commission test site in Nevada (Kaimal and Haugen, 1975) . 90 I6r / 14 - 12 /. 10 Y V RAD g ' (m/seq /'.•;•■ h=122m 6 >' /»* ' • 4 < 2 111' i -6 -4 -2. — -<* ■_ ' 2 4 6 8 10 12 14 16 ■•/■■ ;■:.?. -2 ANEM , • / • (m/sec) • 4 - >" _ 6 L V RAD (m/sec) •. -V • -2 h=148m ./ /. & ■ 2 4 6 10 12 14 16 V ANEM (m/sec) Figure 15.4. Overall comparison be- tween XONDAR-observed wind speeds (Vrad) anc ' those reported by an anemometer (V anem ) mounted at the 122-m level (.upper) and 148-m level (lower) of a tower near Boulder, Colorado, in 1975. In 1975, a prototype, two-component (wind only) XONDAR was compared against instruments on NOAA's 150-m tower at Boulder, Colorado. Figure 15.4 shows the results of comparisons made by NOAA between the XONDAR and conventional anemometers located at two levels on the tower. Each point in Fig. 15.4 represents a 10-min average speed from both the XONDAR and the corresponding anemometer. Model 300 and Model 600 XONDARs have recently undergone a series of comparisons against tower-mounted anemometers in Japan. Preliminary comparisons were made between a 4-kHz Model 300 system and anemometers mounted on a 213-m tower in Tsukuba, Japan, in December 1977. Figures 15.5 and 15.6 show the comparison results. 91 16 - 14 - ■a c o 12 - r/l U O H Pvv P. 10 - \ en \ U v * CD •VsA 1 \\ / / \ V/ CD / j B 8 - \'\ o •o 0) CD i / '•J V\/ 73 6 - \ \V 1 V If V c 4 - nv/i r> a 1 v( / \ i H tt 2 - - 1 1 i 1 1 1 Time of Day N - 360 - 1600 1700 l 1 1800 1 1900 1 2000 2100 i i J& ^^-\ x> <£ST \ /Q-^C /Jr* 300 - Yr^ 1 w^& ^h^^ W - A 01 0) u 240 - m 1) •a X X Anemometer (3 O O O XONDAR o S - 180 - Hi >H 5 T3 a > ** 120 - E - 60 - N - - Figure 15.5. Comparison of wind speed and wind direction measure- ments from a Model 300 XONDAR and an anemometer mounted at the 100-m level of the 213-m high Meteorological Research Institute (MRI) tower, Tsukuba, Japan, in December 1977. 92 1 — — t~ —i— — i 1 r Time of Day 1600 1700 1800 1900 2000 2100 N - 360 1 1 1 1 L 300 W - 240 c o « S - 180 - 120 60 &&* ~<\, A or •4$r&^#-#r* c-X X X Anemometer O O XONDAR 14 Figure 15.6. Comparison of wind speed and wind direction measure- ments from between a Model 300 XONDAR and an anemometer mounted at the 150-m level of the MR1 tower in December 1977. 93 iao 22s ■ 270 ■315 360 < AV= 75 » ( H - 76 « IS ib ! p) CO O u o< n .-3 s ji l/l "T 1 > ift <^ CO i/l n sO in CO *o i— — -U UJ i UJ u fU (/) i + + + + ♦ HJ u, + KJ CO + + — — — — ru vi + ru to + in _ „, z 10 E + + + E CO + + + + i > z i + + + + ♦ 1 i -> s E + + 1 ! + + + > + + u UJ co n > s> jo _ UJ a* c- — in Q r- UJ UJ — n f\J _J J ] 1 UJ UJ UJ 3 E3 X -.* eg n -■ T OJ CO CO '.M o UC •I CM in ON r- CJ K -o vO t n u-j ■ T rr CJ CO CO T s «0 CS a o UJ eo OJ ^ 5. (9 Q o - - t* r^- vO Q 00 M r~ cu > o» — •— CO c> -JO DO CO Q* w ca CO « v> r» j- e» CJ — CM CO UJ «1 p« r- » s s eg tt) UJ to r- CD n _ „ OJ ui to o . ■*- + + + s + + + + r* X s. + + c~ li. s + + + r~ i> U. s r** a sT " -&J + + <7 "■ 2 + + •sT ~ UJ E 1 + + + + ■sT ~ m £ ♦ + + N - V) E + + + * N OJ M + + + + + OJ r— OJ r— w r- 1 OJ r^ OJ 1 s. V ts x CJ s ■-a sO sO cs „ cn cr "-> „ CO or Q „ to ex CS ii s in a. CS OT Ul ca H o ea ;» s cs as (- u c_j CS CS ea ca ■s h- u sa S ra rj -2 a UJ -j -J :i a 1 h> <*>: m ca in CS Ul UJ ^ I— m o u'. ra VI s UJ iii j >r- in UJ _1 CO CO «T < £ <3 5. st E *~ oj OJ V < ^ « UJ E ™ w OJ CO CO M S UJ - s *■ \ + ♦ ♦ + J> X J N. > s S E i E ' s E , s i i ^ : 1 in to 1 to ! I U) 1 en to UJ UJ m o> ^ sO ra UJ Ul m n vO SO in UJ UJ — — <7 ■vT CO CS Ul Ul a^ o sU C> CO Ul u eg — Q s T UJ u) OJ ■a Q o o »n Ji i,o in in vO >n in in in •4J 7" T ^ in in U) in in 7* !•) •T -T n vO in n CJ r* CO ^ CS _ Q CJ O o K) co r- r- Q CJ en _ CO sO to a U o in cs r* C3 UJ *. UJ UJ CS UJ ..J •- Ul Ul :. Ul Ul CS UJ ^ .. UJ VI •o -o >c ■o r*. NC- sO sO r- + r- o a. \ + + + + + + r» CO tt. X + + + + + + r^ c 3u V, + + + + + + \ - yj r v V) r to V (/) S cn T N tM U) T vO .. O vO vO sO sO OJ 00 1 OJ m OJ rn i eg m 1 OJ m OJ m s ■\ s N, o sO *o ! sU sO S s .. to cr 11 „ to II „ c/l a: -J „ CS II . to H Q S ES s s> '-. H UJ n « C5 CS ea h- UJ n CS *^ ca SJ <■; !- UJ 53 -j n CS CS IS H Ul s S pa CS n PS f- UJ ca cs C? E9 O C W UJ I h- z <. H JO q in q in % UJ UJ ! f- in "j m ts JO "- UJ U] ) I- m ca m -j in r j UJ 1 i- Ul ea in cs to « UJ UJ : h* m ea in cs j> CS UJ t- «n cs uo CS m o — OJ 01 CO CO ca sT n Si CS ca \ u eg sc CO O Si T o OJ OJ sr rr \A ts u UJ UJ Ul ry w Si eg eg ■: :j — fi) en _- — OJ ■M i i ru m i i — _ — 5 en -~ — i ■— — — nj l/l i '1 ' 1 1 — — s to 1 i i 1 — OJ > N + + + + + + -> N. ♦■ + + + -> N. ■ 1 \ i 1 1 i -> s i 1 1 1 E 2 1 j: E 1 s i: in ' m to m 1 « to u. UJ n eg ^ p« -u UJ UJ JO m CO M Sf UJ uj in sO T r- TN 'I UJ Ul — CO CO m CO r- UJ Ul r* SJ T) sO no t> UJ UJ JO 9 CO CO r— — 1 Ul J UJ ! 1 Ul .1 UJ 1 UJ CO ir - ■"i n n s .'■ IT j » in ^r -'l sO 'j CtL CO CM r* Tv eg Ul It ca sr r- CO ue eg OJ si m o IE OJ CO s ? j *J o OJ sT CO s LJ « UJ -u CS UJ . ,-B Ul Uj (S UJ Ul iz Ul UJ CD ~ UJ m > OS in rr S m UJ yi ns OJ ea Ul t/J CS ea S3 OJ CO CM ro Ul in CO OJ a m UJ in CO CJ Cj CO UJ t/J Cs CJ ca CO CO f^- CO a. V, + + «■ + — — c* CS Q. \ + — + + — ^ OJ 0. X — — — — — — r^ 1 n. V — — + — — — r~ o X s p* CO X S\ + s. s w V + + N to y + + + S U) E ♦ + + + + + ■s i/i > + + + + + S to T. + + + + + + s l/l + + + + + ■T .. sT si ST ^ ^r eg o OJ T ^ OJ CO CM n cu to OJ CO N C4 N OJ s OJ CO — a cu r— 1 > - — * CO CD u U — ' lH ■H 01 4-1 *J — 1 i- cn s_^ o « ^ T — OJ — X! u CO +J o cn : u- M ♦J >H aj a i— l > u2 ra (U u u — — •• T3 CO ex ■H S-i & +J -^ eo CO ■H O Ou z 1) OJ 01 a -i d ^r 01 >^ 1j CJ +j u w< •H •H U S o L7- r— U QJ 01 ^3 E > cfl -U' H 4*1 Cfl Cfl ■a o — d H CO 4-1 OS ^J < *J 01 CJ 0= > § 00 o — — X 01 S-i ja cfl ■^ s o UJ. Cu u Jj C/l ■H u 3 4-1 ^ d a. i — i 0) E cc 4J 03 cn in m u •r-l 01 in cu > d e 4-> 1 »H o 01 U 2; > O — 1 i— 1 ^"s 01 01 4-1 3 > ■d H 00 ^H 01 in > r-l 01 01 Jul 01 > 3 -C Oj 96 15.5 REFERENCES Balser, M. (1976): Measuring wind shear. In Airports International , IPC Business Press, Surrey, England. Balser, M., C. A. McNary, and A. E. Nagy (1974): Acoustic backscatter radar system for tracking aircraft trailing vortices. J. Aircraft 11:556-562. Balser, M., C. A. McNary, and D. Anderson (1976a): A remote acoustic wind sensor for airport approaches. J. Appl. Meteorol. 15:665-668. Balser, M., C. A. McNary, A. E. Nagy, R. Loveland, and D. Dickson (1976b): Remote wind sensing by acoustic radar. J. Appl. Meteorol. 15:50-58. Kaimal, J. C., and D. A. Haugen (1975): Evaluation of an acoustic Doppler radar for measur- ing winds in the lower atmosphere. Proc. 16th Radar Meteorol. Conf., 22-24 April 1975, Houston, Tex., American Meteorological Society, Boston, Mass., p. 312. Kaimal, J. C. , and J. W. Wescott (1976): An acoustic Doppler sounder for measuring wind profiles in the lower boundary layer. Proc. 17th Radar Meteorol. Conf., 26-29 October 1976, Seattle, Wash., American Meteorological Society, Boston, Mass., pp. 290-296. Nishinomiya, S., and Y. Akai (1979): Wind profile measurements with a Doppler-acoustic remote sensor in the lower atmosphere. Central Research Institute for Electric Power Industry, Iwato 1, 229, Komae City, Tokyo, Japan (in Japanese), 28 pp. Noonkester, V. R. (1978): Multi-sensor measurements of ocean based convective activity. Proc. 18th Radar Meteorol. Conf., 28-30 March 1978, Atlanta, Ga . , American Meteorolog- ical Society, Boston, Mass., pp. 55-64. 97 16. GMD-1 RADIO WIND SOUNDING SYSTEM Robert B. McBeth National Center for Atmospheric Research Boulder, Colorado, U.S.A. 16.1 PURPOSE AND USE The GMD-1 rawin set is a transportable radio direction finder that automatically tracks a balloon-borne radiosonde transmitter. A radio signal containing meteorological information in the form of amplitude or frequency modulation is received, amplified, and detected by this equipment. The detected radiosonde signal is passed to separate equipment in the system where it is recorded. By reference to calibration data, these recorded data are converted to values of temperature, humidity, and pressure. Recordings of time versus progressive changes of the elevation and azimuth positions of the ascending balloon, as determined by tracking of the signal from the radiosonde, are converted to wind speed and direction . 16.2 SYSTEM COMPONENTS 16.2.1 Radiosonde VIZ 1680 MHz This equipment consists of a transmitter, a modulator, an antenna, a battery, and pressure-, temperature-, and humidity-sensing elements. The pressure capsule employed is temperature-compensated and constructed to give approximately logarithmic response over the operating range. The temperature element is a rod of ceramic material with a high negative resistance coefficient. The humidity sensor is a plastic strip with a gelatinous cellulose coating that contains finely divided carbon particles in suspension. Resistance of the sensor increases with relative humidity. The radiosonde in use for the low-level experiment was the VIZ 1680-MHz radio- sonde equipped with ACCU-LOK sensors. These sensors are calibrated at the factory, thus eliminating the need for a baseline check before each observation. The assembled instru- ment weighs about 800 g and can be carried to an altitude of 30 km by a helium-filled balloon. The battery furnishes power to the transmitter and modulator. The transmitter operates in the 1660- to 1700-MHz band, and its carrier is amplitude-modulated by an audiofrequency pulse, the rate of which is determined by the pressure-, temperature-, and humidity-sensing elements. Switching between temperature, relative humidity, and a calibration signal (reference) is achieved by a pressure-operated device called a baroswitch. With the ascent of the radiosonde the expansion of an aneroid capsule, through connecting mechanical linkage, causes a contacting pen to travel across a printed circuit board array of silver conducting strips mounted on insulated material. This is termed the commutator. When the baroswitch pen rests on an insulated portion of the commutator the value of air temperature is transmitted by the radiosonde unit. The silver segments of the circuit board are used to switch the sonde to reports of relative humidity or references. The first four of every five silver segments on the commutator switch the radiosonde to report values of relative humidity whereas the fifth, tenth, fifteenth, etc., silver seg- ments direct the sonde to a circuit which provides a fixed frequency of about 190 Hz. When this frequency is received at the ground station the meteorological recorder pen moves to approximately the 95th ordinate of the recording chart. This provides the radiosonde operator with a reference that enables him to adjust the setting of the recorder to exactly 95 ordinates and thus remove any subsequent linear errors in the temperature and 98 humidity values to be received until the next reference. On most U.S. radiosonde devices the 30th, 45th, 60th, etc., contact reports a slightly greater frequency, termed the high reference, which is used to facilitate counting the number of contacts received at the ground station. Every radiosonde unit is delivered with a calibration chart which provides a value of pressure for each contact of the particular sonde. 16.2.2 Rawin Set The rawin set automatically tracks by continuous homing on the radiosonde signal, The equipment indicates and records the azimuth and elevation angles of the radiosonde. These angles are plotted with the height (computed from the pressure and temperature data) to determine wind direction and speed. 16.2.3 Recording Equipment Radiosonde meteorological data (temperature, humidity, reference data) are recorded on a time-frequency strip chart recorder. Elevation, azimuth angles, and elapsed time after balloon release are printed on paper tape. 16.3 TECHNICAL CHARACTERISTICS Power input to the system is 105-129 V a.c., 50-65 Hz, 1,000 W. The frequency range is 1660-1700 MHz, and either an AM or FM signal can be used. The RF system features conical scanning with a single dipole antenna and a parabolic reflector. The receiving system uses a superheterodyne receiver and an IF frequency of 30 MHz; its tracking accuracy is 0.05° maximum error between 10° and 60° elevation . The antenna positioning system features automatic tracking, with the option of manual control locally. 16.4 DATA PROCESSING At most operational GMD radiosonde stations data are manually processed. Temper- ature and humidity strip chart records and printed elevation and azimuth angles from the control recorder are analyzed with the aid of slide rules, graphs, and scales to yield profiles of temperature, humidity, and wind. For the BLIE, with six scheduled soundings per day (two of these on the tower carriage), there was not enough time for manual process- ing. Instead the operator manually transcribed data from the two hard-copy records onto a magnetic tape from which the data were entered by phone line into a Bureau of Reclamation computer. When conventional radiosonde reduction programs were used, the computer yielded processed data on a printer at the field site with a normal turnaround time of 1 min. 99 17. TDFS LOW-LEVEL RADIOSONDE SYSTEM C. Fink and E. Schollmann Upper-Air Research Station Deutscher Wetterdienst Munich, Federal Republic of Germany A. Kb'lbl Instrument Division Deutscher Wetterdienst Munich, Federal Republic of Germany 17.1 LOW-LEVEL SONDE TDFS The low-level sonde TDFS is a lightweight weather sonde for measuring tempera- ture, humidity, and atmospheric pressure in the free atmosphere up to about 600 mb . For dry- and wet-bulb temperatures, bead thermistors with negative temperature coefficient (NTC) resistance, having diameters of 0.4 mm, are used. The measuring range is from -40°C to +40°C. The frequency of a subcarrier oscillator is changed in the range of 300-900 Hz by the temperature behavior of the NTC resistance. This subcarrier frequency modulates a crystal-controlled transmitter with a frequency between 402 and 406 MHz. The exact fre- quency can be determined subsequently for each sonde by a plug-in crystal. The modulated signal is transmitted by a half-wavelength antenna, which also serves as a means for sus- pension. Two small 9-V dry batteries are used for the power supply. The pressure- measuring element is an aneroid capsule of copper beryllium, which is temperature com- pensated over the entire measuring range. The measuring range of about 1,050 mb to about 600 mb is subdivided by gold-plated contacts in pressure steps of about 25 mb . The features of the sonde are listed in Table 17.1. The case is made of white plastic that insulates well against cold. Its dimen- sions are 206 x 170 x 85 mm; the weight including batteries is 300 g. Both temperature sensors are housed in a foamed plastic tube that protects the sensors from mechanical influences and direct solar radiation and provides for an optimal air flow. Three views of the sonde are shown in Fig. 17.1. 17.2 RECEIVING STATION KS 75 The signals of the radiosonde are received by the UHF radiosonde receiver. Here the demodulation takes place; i.e., the measuring signals (frequencies of 300-900 Hz) are separated from the high frequency carrier and are available for further processing at the receiver output. The special data-handling unit allows the low frequency signals of the sondes to be automatically evaluated. For this purpose four coefficients have to be inserted to describe the corresponding calibration curve. On the basis of the inserted sonde calibra- tion curve, a microprocessor calculates the digitally measured low frequency values and shows a temperature in degrees Celsius in a digital display, at 1-s intervals. During sonde reception the dry- and wet-bulb temperatures are received alter- nately, interrupted at irregular intervals by pressure signals that are represented on the strip chart recorder as rectangular steps. Figure 17.2 illustrates a portion of a typical strip chart record. Temperature values of -50°C to +50°C can be recorded with a resolution of 0.4°C/mm. Two digital outputs (TTL level, BCD 1,2,4,8) available at the back of the receiving station make it possible to connect tape punches (e.g., hp 3,489 A) or computers 100 Table 17.1. Description of the low-level sonde TDFS Feature Description Dimensions Weight with battery Sensors Temperature-measuring range Pressure-measuring element Pressure-measuring range Resolution Accuracy of system Transmitter frequency range Transmitter power Modulation Subcarrier frequency Frequency bandwidth Frequency stability Antenna Polarization Duration of operation Height 206 mm Width 170 mm Depth 85 mm 300 g For dry- and wet-bulb temperature, matched bead thermistors, var- nished in white; diameter 0.4 mm. -40°C to +40°C Aneroid capsule, CuBe 1,050 mb to ~600 mb in pressure steps of ^25-mb intervals; every fifth pressure step is a double step. Pressure Temperature Pressure Temperature Psychrometric difference +0.5 mb +0.1°C + 2 mb +0.4°C +0.2°C 402 to 406 MHz crystal-controlled ^25 mW Frequency modulation 300 to 900 Hz +6 kHz +30 ppm Half-wavelength omnidirectional Vertical ^•45 min with alkali-manganese cells, Mallory MN 1604 for further processing of the data. The features of the receiving station are summarized in Table 17.2. The receiving station and the radiosonde were developed for the Deutscher Wetterdienst by Instrumentenamt Mfinchen. 17.3 EVALUATION Receiving station KS 75 produces an analog record (see Fig. 17.2) containing atmospheric pressure in the form of pressure steps and the dry- and wet-bulb temperature, in the correct physical units. The dry-bulb data are recorded during a period of about 6 s, and the wet-bulb data during a period of about 3 s. From the calibration table attached to each sonde the values of the pressure levels are obtained and the pressure time curve is plotted as shown in Fig. 17.2. Intermediate pressures may be interpolated 101 Figure 17.1. Low-level radiosonde of the Deutscher Wetterdienst : (above, left and right) cover removed to show (a) aneroid capsule, (b) bat- teries, (c) sphere, to create a tur- bulent air flow, (d) blackened air- duct, (e) dry-bulb thermistor, (f) wet-bulb thermistor, and (g) array of pressure contacts; (left) cover on for balloon launch. 102 600 650 700 750 800 850 900 950 1000 -30 -20 -10 10 Pressure (mb) Temperature (°C) 20 30 40 50 Figure 17.2. Analog record of a sounding showing the pressure vs. time curve plotted from the pressure calibration of the sonde and the occurrences of pressure readings. Note the warming to 0°C of the wet bulb at the instant of freezing. from this curve. Then the significant levels of the dry- and wet-bulb temperature are selected. Pressure and dry- and wet-bulb temperatures for each significant level and for any mandatory pressure levels are inserted in a small programmable calculator to yield relative humidity, dew point, and height. The data for mandatory height levels are deter- mined by interpolation. If a calculator is not used, the relative humidity has to be determined from psychrometric tables. The computation of the height is then made by using the Stiive diagram. 103 Table 17.2. Description of receiving station KS 75 Feature Description Dimensions Weight Power input Input of calibration curve Temperatures Analog recording Height 440 mm Width 520 mm Depth 400 mm without paper cas- settes and dust cover; 490 mm with cassettes and cover. 40 kg 220 V + 10%, 50 Hz + 20%, 100 VA 4 coefficients Digital indication Alternating between dry- and wet- bulb temperatures, interrupted by pressure steps; temperature range -50°C to +50°C. 17.4 ACCURACY Each sonde is calibrated by the manufacturer. The conversion of frequency into temperature is made by means of a polynomial of the third degree. Since only the coeffi- cients of one calibration curve can be inserted into the receiving station KS 75, the therm- istor for dry- and wet-bulb temperatures have to be matched. A deviation of +0.1°C is acceptable. Because of digitalizing , the resolution of the station amounts to 0.1°C. When all sources of error are combined, the system accuracies are +0 . 4°C for temperature and +0.2°C for psychrometric difference. The lag-coefficient of the bead thermistors is about 2s; it is larger with ice accretion. The change of the water from liquid to solid on the wet-bulb thermometer takes about 30 to 60 s, depending on the degree of supercooling and the moisture content of the air. When a pressure step is reached during the ascent, a fixed resistor is switched in instead of the temperature sensor. The resulting step change in frequency provides an indication of the pressure contact on the strip chart. When the points are connected it yields the pressure versus time curve in Fig. 17.2. The leading edge is chosen for the calibration value. The calibration accuracy is +0.5 mb . The occurrence of the event is reported with a maximum delay of 0.5 s. The accuracy of the entire system is +2 mb . 104 18. CORA RADIOSONDE SYSTEM USING FREE-FLYING BALLOONS Ilkka Ikonen Vaisala Oy Helsinki 42, Finland CORA is an automatic upper-air system commercially available from Vaisala. The system is equipped with automatic computation and analog output for monitoring purposes. Manual evaluation is possible by using the analog output of radiosonde frequencies. The program for upper-air sounding stations, which make soundings for meteorological networks, is called CORA 6. In the BLIE a special computer program called CORA 8 EXP was used. The radiosonde used by the system is the Vaisala RS 21-12 CN, shown in Fig. 18.1. 18.1 RADIOSONDE RS 21-12 CN 18.1.1 General Characteristics The general characteristics of radiosonde 21-12 CN are as follows Output frequency: 46.1 to 50.8 kHz. Carrier frequency: 403 MHz (adjustable +3 MHz). Sensors: Low-altitude pressure P, aneroid barometer. High-altitude pressure PP (above 150 mb) , aneroid barometer. Humidity U, Humicap thin film capacitor. Temperature T, NiFe-alloy bimetal. Two reference capacitors Kl and K2. VLF receiver for the Omega system (13.6 kHz). Figure 18.1 Vaisala radiosonde RS 21-12 CN before launch during BLIE. 105 The pressure and temperature sensors are linked mechanically to variable capaci- tors. Each of the six elements is connected in sequence to the modulating circuit by a switch attached to a rotating reel. The sonde is tied to the balloon by a line running from the reel. During ascent the weight of the sonde pulls line from the reel, causing it to rotate. During one cycle each element transmits for about 1.6 s. Winds are derived from the VLF reception of Omega signals. 18.1.2 Calibration Each radiosonde is individually calibrated in the factory. Numeric calibration coefficients for P, PP, and T are available for automatic computation. U calibration is carried out during baselining. 18.2 GROUND EQUIPMENT Hardware consists of the following main parts: Teletype 43 ASR . NOVA 2 minicomputer with 64-Kbyte memory. NOVA cassette (3-drive C-cassette) . 400-MHz receiver. Omega receiver. Radiosonde signal sampling unit. Antennas for 400-MHz and local Omega reception. Analog output devices for monitoring and manual evaluation. Baselining equipment. 18.3 AUTOMATIC COMPUTATION 18.3.1 Output Format The system offers two possibilities for data output: teletypewriter and C-cassette. Output made on the teletype has the following format: tttt ZZZZZ PPPPP +TTT UU DDD FFF , where tttt is time for the level in seconds; ZZZZZ is height in geopotential meters computed from pressure, temperature, and humidity using the hydrostatic equation; PPPPP is pressure in 0.1 mb ; TTT is temperature in 0.1°C; UI is relative humidity in %; 1)1)1) is wind direction in degrees; and FFF is wind speed in 0.1 m/s. The system offers a wide variety of post-ascent processing capabilities using the data stored on the magnetic cassettes. Programs can be written and run in BASIC language using the CORA equipment. 106 18.3.2 Automatic Computation of PTU Radiosonde signals from the six elements are received continuously, six fre- quencies in sequence per data frame. Each of the six frequencies is passed through its own digital filter that rejects impossible values. Meteorological values are computed using the following procedure: (1) Compute the corrected element value for P, PP , T, and U using the element in question and both reference capacitors. (2) Compute the humidity value using a predefined calibration equation that is fixed during baselining. (3) Compute pressure and temperature using individual calibration equations of second degree. Pressure, humidity, and temperature at a specified time are computed as weighted averages of the two nearest observed values. Weighting is in proportion to nearness to the speci- fied time. 18.3.3 Manual Computation of PT U Manual computation is easily carried out by an operator using the automatic radiosonde receiver and evaluation ruler. 18.3.4 Wind Computation Specification Wind computation is based on the Omega network. All eight Omega stations are always used. Both local wind and remote wind are determined, and a differential cor- rection is made optionally. Wind is reported each 10-s interval, which corresponds to a full Omega sequence. In derivative computation a sliding second-degree curve across 4 min is used. A weighting is carried out to get an optimal solution in the sense of least squares fitting. If the quality is under a certain pregiven level, the weight of that particular signal is zero. Although the accuracy of Omega-derived winds depends on local Omega reception conditions, it is the experience of Vaisala that, in most cases, the wind vector is accurate within 1-2 m/s. 18.4 INTERCOMPARISON DATA (1) Automatic computation listing in the format described in 18.3.1, at 10-s in- tervals . (2) Pressure, temperature, and humidity transferred on the Vaisala low-level sound- ing aerogram. (3) Manual readings from automatic radiosonde receiver, transferred on the Vaisala low-level sounding aerogram. 107 19. THE AIRSONDE SYSTEM David B. Call Atmospheric Instrumentation Research, Inc. 220 Central Ave. Boulder, Colorado, U.S.A. Alvin L. Morris Ambient Analysis, Inc. 3300 Arapahoe Ave. Boulder, Colorado, U.S.A. 19.1 INTRODUCTION The Airsonde™ meteorological sounding system (Fig. 19.1) is a complete, highly portable system in which precision sensors and solid-state electronics are combined to yield digital data in standard meteorological units in real time. It is designed for use with small (30- and 100-g) balloons and is capable of reaching heights greater than 10 km. The Airsonde sensor package (hereafter called the Airsonde) is contained in an expanded polystyrene package (see Fig. 19.2) in the form of a helicoid propeller. As it rises or descends, its spinning aspirates two bead thermistors mounted in radiation shields at the tips of the propeller. One is covered by a wick that is wetted by water contained in a small reservoir, and the two form a psychrometer . Pressure is sensed by an aneroid capsule whose temperature is measured by a third bead thermistor. No baseline measurements are required for most uses of the Airsonde system. Figure 19.1. The Airsonde system with an inflated 30-g balloon. The aluminum case to the left contains the ground station. 108 Figure 19.2. Schematic diagram of an Airsonde . Figure 19.3. Airsonde ground station mounted in aluminum suitcase. Modified HP-97 is shown in foam cutout. Similar cutout is for tape recorder beside Airsonde in front of case. An electronic multiplexer samples the sensors sequentially once every 6 s. The Airsonde mass, complete with battery, is 130 g. The transmitter, a narrow-band, crystal- controlled, solid-state device, transmits data in analog form, radiating 25 mW of power at a frequency of 403.5 MHz. The Airsonde ground station (Fig. 19.3) is interchangeable with that of the Tethersonde, described in Chapter 20. Enclosed in an aluminum case, the ground station consists of a receiver, a small strip chart recorder, a microcomputer, a visual digital display, a battery charger, and a power supply. It may be operated on either 110 V a.c. or 12 V d.c. For soundings up to 300 mb the ground station may be used with an omnidirec- tional, half-wave antenna; for higher-level soundings a higher-gain antenna is used. In the ground station the incoming analog telemetry signal is converted to a digital signal, which is then processed by the microcomputer. Digital data from the ground station may be printed on a modified Hewlett-Packard Model HP-97 programmable printing calculator, or recorded on either digital or analog tape recorders. Table 19.1 shows sample printout information from an HP-97 during an Airsonde flight. Figure 19.4 is a graph of the Airsonde sounding data from a flight. 19.2 OPERATIONAL CHARACTERISTICS The Airsonde used in BLIE measured pressure, temperature, and wet-bulb tempera- ture. Complete profiles of these variables, or of variables such as relative humidity or potential temperature that can be derived from the measured variables, are attainable. 109 Table 19.1. Examples of Airsonde system data printouts from a modified HP-97 printing calculator Observed data printout"-' Observed and calculated data printout! Variable^ Printout Variable§ Printout Time T (°C) Tw(°C) P(mb) 20.8 15.4 8.6 820.4 Time P (mb) T(°C) RH (%) w(g/kg) z(m) 6(K) 9.2139 818.6 9.0 43.8 6.0 17.6 305.6 * A real-time printout of observed data only. The time, 20.8, is elapsed time in decimal minutes after the ground station was set to receive and pro- cess data. Real-time printout is for every second or third frame of data. t A printout of observed and calculated data from post-flight playback of tape recorded data. The time is read 9 h, 21 min, and 39 s. On playback every data frame can be printed out if desired. § Note that the variable columns shown here are not part of the HP-97 printout . The basic Airsonde ground station is quite versatile because it has both analog and digital outputs and because it interfaces with a number of peripheral devices. Commu- nication with these devices is through two 25-pin connectors, one of which provides RS-232C and 20-mA signal levels to operate a cathode ray terminal, teletypewriter, or modified HP-97 printing calculator, whereas the other provides a terminus of 16 bi-directional data lines, 8 of which are control lines. The latter connector gives the user a general-purpose programmable interface and provides control and data signals as well as power for recording data in ASCII FSK on an analog cassette tape recorder. The most commonly used peripherals are a magnetic tape recorder and a modified HP-97 calculator. Digital data are tape recorded during flight. In addition the HP-97 may be used to calculate data shown in Table 19.1 (right) for every fifth frame or to print every third frame in real time. After the flight the taped record may be played back through the ground station to the HP-97 at a selectable slower rate, and the HP-97 may be programmed to do various calculations with every data frame, printing calculated as well as measured variables. The program has certain editing features. For example, while integrating the hydrostatic equation to calculate height, the program compares current and previous frames of data for apparent discrepancies. If the difference in either temperature or pressure is excessive according to the criteria programmed into the calculator, the program rejects the current data for calculations, but prints current time, pressure, and temperature. The operator can observe these and override the program decision at will. The Airsonde is normally flown on either a 30- or 100-g balloon. Guidelines for balloon inflation are provided by the curves of Fig. 19.5, which relate ascent rate of the Airsonde to weigh-off mass. (Weigh-off is the total mass, including inflation device, suspended from the balloon during inflation. Gas is added until the balloon is in neutral equilibrium with the weigh-off mass.) Separate curves are given for both helium and hydro- gen and for 30- and 100-g balloons. The curves are approximate and are based on data ob- tained by dropping the Airsonde from a tower and considerations of balloon drag and buoy- ancy. An ascent rate of more than 1 m/s is necessary to assure adequate aspiration of the psychrometer . In the intercomparison experiment the Airsonde was flown in tandem with other types of radiosondes from a single 300-g balloon. 110 Airsonde Sounding-Marshall, Co., 0921 MST, 29 Apr. 1978 250 50 Q 300 © x^_ ©o ^r x n — i — 100 i 1 1 r si E 400 500 600 700 800 900 1000 © ©^ X J Q © °© T % 5 ©, ©, /'Wet bulb freezing-Rdg 0.1° C © ^ *x x*' x J ± ± 5 (m/s) V, -45 -40 -35 -30 -25 -20 -15 -10 -5 5 10 15 Temperature (°C) 250 300 400 500 600 - 700 800 900 1000 Figure 19.4. Sample Airsonde sounding, every fifth data frame, with addi- tional detail between 750-760 mb and 550-560 mb . Note that the wet-bulb freezing temperature provides a convenient temperature check. The HP-97 program used for this sounding calculates vapor pressure over ice for all values of T <0°C. 19.3 SENSOR CHARACTERISTICS 19.3.1 Tempera ture-Hujnidity The precision bead thermistors used in the Airsonde are epoxy-coated sensors manufactured to precise tolerances. Interchangeability and accuracy of +0.2°C from +50° to -20°C are stated by the manufacturer; typical performance is better. The Airsonde system ground station measures the resistance of each thermistor and interpolates data from a table to derive dry-bulb and wet-bulb temperatures. Error intro- duced by the mathematical computation is an insignificant 6.0 x 10 3 °C in the worst case. Since the psychrometric technique of measuring humidity is more sensitive to wet- bulb depression than to either dry- or wet-bulb temperature, the circuitry was designed to assure an accurate measure of the depression. An electronic multiplexer switches the dry- and wet-bulb thermistors into the measurement circuitry within 1 s of each other during each 6-s frame. This is considerably faster than an Assmann psychrometer can be read. A possible 0.1°C error, traceable to manufacturing tolerances in reference resistors and to 111 600 500 400 "5 300 0) 200 100 I ■ I • I ■ T. hi - /// ///} rl! — Hydrogen //// I ! ii " hi - /// ! I _ I 100 g //// \^J / // Wfi / ; M Aa< ///A ///.\ \ /// \\ A" \\ sr/ ;;„ — sZrs 30 g y^f^ ^^ ^r ^^^ V z (ft/min) (X10 2 ) - 12 3 4 5 6 7 I I I I I I I r 1 — | i 1 i 1 i' i 1 i 1 _ 2 V z (m/s) Figure 19.5. Graphic aid for calculating the weigh-off mass to be used when inflat- ing Airsonde balloons. switch resistance, is common to both dry- and wet-bulb temperatures and is therefore absent from the wet-bulb depression. Good psychrometry requires that the sensors be properly exposed. They must be adequately aspirated and protected from the effects of radiation. Fast response is essen- tial in a sounding system that moves rapidly through strata in which humidity gradients are large. The dry-bulb thermistor has a time constant of 10 s in still air and 3 to 5 s with the rotational aspiration that occurs at nominal ascent rates (2 to 4 m/s). The wet-bulb response time, which has been observed only in a very limited way, is believed to be about three times that of the dry bulb. Each thermistor is mounted in its own radiation shield. The shield, a tube of molded expanded polystyrene, has small thermal mass and is a good insulator. Since the Airsonde rotates about a vertical axis during flight, direct insolation to the bead can occur for only a small fraction of each rotation and is therefore not significant. Diffuse radiation is minimized by blackening the inside surfaces of the shields. Aspiration is achieved by the spinning of the helicoid propeller-shaped Airsonde as it ascends or descends. A minimum aspiration rate of 3 m/s is recommended (Bindon, 1965). The Airsonde is designed to produce an airspeed past the propeller tips equal to 112 three times its ascent rate. The helicoid shape also assures that airflow over the entire propeller is parallel to the surface. Thus, the air flows axially through the radiation shield tubes at the propeller tips. 19.3.2 Pressure The Airsonde pressure transducer is a variable capacitance aneroid cell consist- ing of a square ceramic substrate with the aneroid capsule bonded symmetrically to both sides. A metalized area of the ceramic substrate forms one plate, and the capsule forms the other plate of the sensing capacitor. Hence, the sensing capacitor is inside the cell where the dielectric is an unchanging vacuum, resulting in a small and rugged sensor. A similar pressure sensor has been in use at NCAR for several years (Pike and Bargen, 1976). The cell is 3.18 cm on a side and 0.76 cm thick. It has a capacitance of 10 pF at 1,000 mb . Its sensitivity is 0.03 pF/mb. Unlike the dry- and wet-bulb thermistors, whose calibration and interchangeabil- ity are well defined, each aneroid cell must be calibrated individually. The calibration yields a curve of pressure versus capacitance that can be fitted adequately by a second- degree polynomial. Four coefficients, one of which is a temperature compensation coeffi- cient, are provided with each Airsonde. These are placed in the computer through thumb- wheel switches before launch. The computer executes a program that does the curve fitting in real time for every frame of Airsonde data. This technique gives pressure readings in millibars with a characteristic absolute accuracy of +3 mb from 1,000 to 300 mb . Accura- cies of +1 mb can be achieved through that pressure range if a baseline correction from a good reference barometer is entered into the ground station. Under standard procedures the Airsonde is not calibrated for pressures less than 300 mb . Hence, flights above 300 mb may have errors that are considerably greater than +3 mb , unless the Airsonde has been cali- brated at those pressures and an additional correction is made to the data. 19.4 ELECTRONICS 19.4.1 Airsonde Figure 19.6 is a block diagram of the Airsonde. The transmitter is a conven- tional frequency-modulated crystal-controlled circuit. It uses a voltage-controlled crys- tal oscillator and 9X frequency multiplication to generate 25 mW at 403.5 MHz. Temperature stability and narrow +5-kHz frequency deviation permit use of narrow band receivers with very high sensitivity. The combination of small transmitter power and high receiver sensi- tivity allows a standard 9-V transistor radio battery to serve as the transmitter power source. This battery is widely available, lightweight, sealed, and inexpensive. Sensor conditioning circuitry is simple, requires little power, and uses inexpen- sive, readily available components. The use of precision references for both resistive and capacitive sensors eliminates the need for calibration of each sonde. The Airsonde has two RC data oscillators. One provides a frequency whose period is linearly proportional to resistance. An electronic switch acts as a timing element and inserts three reference resistors and then three sensing thermistors sequentially in this RC oscillator. The first resistor (Rsync) uas a ^ ow va l ue that establishes a high synchro- nization frequency, which the microcomputer interprets as the start of a frame. The next two reference resistors (RloreF anc ' R HIREF^ are P rec isi° n components whose accuracy allows the microcomputer to calibrate che data link in software during each data frame. Knowing the values of these two reference resistors (10 kft and 100 kfi) , the microcomputer can measure the resistance of three subsequent thermistors (R^t^T' ^TDRY' anc ' RTWET) t0 an accuracy of 1%, from which it can compute temperature to ±0.2°C, the inherent thermistor accuracy. The circuit has particular advantages in psychrometry because both dry- and wet- bulb sensors use exactly the same measurement circuitry separated in time by 1 s . The only error sources in the circuitry that can contribute to a differential error are switch re- sistance differences and digitalization errors in data processing. The electronic switches 113 "SYNC ;, -wv ^a_ 'LO REF -vw 'HI REF — vw TINT — w^ — 'T DRY -vw 'TWET REF ^PRES V RC OSC. Frame OSC. Antenna Control Logic 403.5 M Hz XMTR RC OSC. Crystal Figure 19.6. Block diagram of an Airsonde. are contained in a single, mass produced, integrated circuit. The typical ON resistance of these is matched within a few ohms, an insignificant difference. A second RC oscillator uses a CMOS gate as the frequency source. It switches sequentially first to a stable reference capacitor (Cppp) and then to the aneroid pressure- sensitive capacitor (Cpppc) . The ratio of these two frequencies is proportional to pres- sure. The thermistor that measures internal Airsonde temperature (RtINT^ i s mounted on the pressure sensor, and its temperature is used by the microcomputer to correct the pressure measurement for the temperature coefficient of capacitance of the aneroid cell. 19.4.2 Ground Station The Airsonde-Tethersonde ground station (Fig. 19.3) is designed for the field environment. Mounted in an aluminum suitcase it is portable and ruggedly constructed. All electronics, including the four receiver modules, are installed on one 7^-in x 9^-in printed circuit card that is mounted on six standoffs to the aluminum front panel. This configuration is easy to service and transport. The ground station can be logically divided into analog and digital sections, as shown by Fig. 19.7. The analog section includes the FM receiver, audio active filters, and phase-locked loops (PLL) . One PLL is tuned for 500 + 125 Hz. This 500-Hz PLL is used with the Tethersonde continuous analog channel and has no function with Airsonde operations. A second PLL, tuned for 2,500 Hz, acts as a bandpass filter with excellent signal-to-noise ratio performance for Airsonde or Tethersonde data. It converts the sine wave signal to a square wave for input to the 16-bit frequency counter. The 8080-A microprocessor and associated memory and peripheral integrated cir- cuits form a general purpose microcomputer. All system software resides in memory where it is available whenever the system power is turned on. 114 Antenna Analog \ / (Q) Output 403.5 MHz FM RCVR Active Filter and LL-2500 Hz Programable Frequency Counter 1 1 F A r— ^ I 16BIT I Data Active Filter and PLL-500 Hz Chart _- RCDR "* ! 8080A Microprocessor A (k Digital Display 199.9 LQ/ B „ ", Analog uutput TTY, CRT or Hand Held Keyboard ZU IVIA Audio Cassette Recorder 9 9 < RS-232C II GP I/O 16 Bidirectional Data Lines ► Switches HP 97 Printer/ Calculator Power Converter -► + 12 V -* — 12 V -► + 5 V ► -► 5 V 115 VAC Figure 19.7. Block diagram of an Airsonde-Tethersonde ground station, The ground station displays and stores Airsonde data in various ways. The prim- ary display is a 3^-digit light-emitting diode (LED) display. All data are displayed in real time. One of seven individual LEDs also lights above an engraved legend that identi- fies the parameter currently being displayed and gives its meteorological units. Hardcopy records can be printed on ASR-33 teletypewriter or HP-97 printing cal- culator (see Table 19.1). The same records can be displayed on most common CRT terminals and line printers. The standard HP-97 calculator has been modified for this application by a small printed circuit card to give the microcomputer access to various storage registers in the calculator. The microcomputer sends data to HP-97 storage registers and initiates execution of programs stored in the programmable HP-97. Permanent storage of data in a form readable by external computers is provided by microcomputer circuitry that controls standard audio magnetic cassette recorders. The Sony TC-142 gives excellent performance and is battery powered. Digital data are converted into a series of discrete audio tones for audio recording. A standard cassette will record several hours of Airsonde or Tethersonde data. These data may be played back into the ground station at any time for conversion to other computer-compatible media. On playback the data are converted to the same format as real-time data, so all peripherals that are used with real-time data will work without change on recorded data. 115 19.5 DATA CONSISTENCY TESTS To determine the consistency of the overall data processing subsystem during flight, the wet-bulb thermistor was replaced by a precision reference resistance equivalent to a 25 . 0°C wet-bulb reading. During the course of a flight from 843 mb to 314 mb , the average wet-bulb temperature reading was 24.94°C, and the standard deviation of the values was 0.05. In a second flight test, the wet-bulb thermistor was left bare so that it was sampling dry-bulb temperature. The two sets of dry-bulb temperatures were compared for temperatures ranging from 27°C at the surface to -5° at 550 mb . The mean difference was 0.08°C, and the standard deviation of the differences was 0.04, which is within the speci- fications for interchangeability of the thermistors. 19.6 REFERENCES Bindon, H. H. (1965): A critical review of tables and charts used in psychrometry . In Humidity and Moisture, Measuremen t and Control in Science and Industry , Vol. 1 , Robert E. Ruskin (ed.), Reinhold, New York, pp. 3-15. Morris, A. L. , D. B. Call, and R. B. McBeth (1975): A small tethered balloon sounding system. Bull. Am. Meteorol. Soc. 56:964-969. Pike, J. M. and D. W. Bargen (1976): The NCAR digital barometer. Bull. Am. Meteorol. Soc. 57:1106-1111. 116 20. THE TETHERSONDE SYSTEM Alvia L. Morris Ambient Analysis, Inc. 3300 Arapahoe Ave. Boulder, Colorado, U.S.A. David B. Call Atmospheric Instrumentation Research, Inc 220 Central Ave. Boulder, Colorado, U.S.A. 20. 1 INTRODUCTION A small tethered balloon system, developed at the National Center for Atmospheric Research (NCAR), has been described by Morris et al. (1975). The Tethersonde^M system, successor to the NCAR system, is similar to it in many ways. Probably the most significant difference to the user is the treatment of data in the ground station. The Tethersonde Model TS-2A described here provides analog data quite similar to those provided by the NCAR system; it also provides digital data in conventional meteorological units. A Tethersonde system ready for flight is shown in Fig. 20.1. The principal components of the system are (1) an airborne sensor package, (2) an aerodynamically shaped balloon (blimp) that carries the sensor package aloft, (3) a variable-speed reversible electric winch used to control tether line, (4) a ground station, and (5) several items of ancillary equipment. The components of a Model TS-2A system, excluding the balloon but including a modified HP-97 printing calculator, are shown in Fig. 20.2. Figure 20.1. Tethersonde system ready for flight. The balloon and sensor package fly better if a rigid spacer (not shown) is placed between the suspension lines at about the level of the lowest part of the bottom tail fin and tied to the tail fin. The properly inflated 3.25 m 3 balloon is about 5 m long. 117 Figure 20.2. Tethersonde system, in- cluding winch, ground station with the sensor package in its cradle, and modified HP-97 printing calculator. The case containing the ground sta- tion is 24 x 47 x 67 cm. The mass of the winch is 20 kg; the mass of the ground station and sensor package is 11.5 kg. During operation, the tethered balloon lifts the sensor package, which samples pressure, temperature, wet-bulb temperature, and wind direction and wind speed in its normal operational mode. With additional, attachable sensors, ozone, carbon monoxide, and the temperature structure parameter Cj have also been measured by the Tethersonde system. Sensed values are transmitted to the ground station through two data channels on FM-FM telemetry using a carrier frequency of 403 MHz. One of the data channels carries continuous data from a selectable sensor; the other uses a time multiplex format to carry data from one to eight sensors. At the ground, the telemetry signal is converted to an analog voltage that may be recorded on a strip chart recorder and to digital data that are displayed by light-emitting diodes. The digital data are also available to external data storage and processing de- vices . 20.2 OPERATIONAL CHARACTERISTICS 20.2. 1 Sensor Pac kage The standard instrument package contains circuits which automatically interrogate the sensors and condition sensor signals for transmission to the ground. Temperature is measured over the range of -50°C to +50°C; relative pressure is measured over the range to 100 mb . Signals are scaled directly for these ranges; signals are also scaled for four 25°C and four 25-mb ranges. By using two range scales for temperature and pressure, ade- quate resolution may be obtained on a small strip chart recorder without causing ambiguity. The telemetry system consists of an FM, crystal-controlled transmitter and re- ceiver. Data are transmitted at 403 MHz in two formats on separate frequency multiplexed audio channels. One channel uses an FM-FM PAM time multiplex format which may be recorded immediately on a single strip chart or audio magnetic tape or both simultaneously. In this format data quality is ensured by including high and low reference data in the recordings. On the second channel, data from any one sensor are transmitted continuously in FM-FM so that spectral information to 10 Hz can be telemetered. A data frame in the time multiplex format consists of a wide sync pulse whose amplitude is full scale, followed by eight sensor channels separated by zero reference values. All sensor data are linear and are scaled in meteorological units for convenient chart interpretation. Standard sensor ranges are the following: 118 Dry-bulb temperature: 25°C and 100°C. Wet-bulb depression: 25°C. Pressure: 25 nib and 100 nib . Wind speed: 20 m/s. Wind direction: 360°. All sensors are calibrated as part of the system using standards which are trace- able to NBS or are based on fundamental physical principles. Periodic recalibration and certification service is available at the factory, but no baseline or reference calibration of any kind is required in the field. It is necessary to have a separate barometric read- ing if absolute pressure is required. 20.2.2 Ground Station The Model TS-2A-GS ground station receives a telemetry signal from the sensor package, processes the signal, and yields both analog and digital outputs. The analog output may be recorded on an external recorder or on a small strip chart recorder contained in the ground station or on both. Digital data are displayed by light-emitting diodes. Two 25-pin connectors permit the transfer of data in several modes. One provides RS-232C and 20-mA signal levels for cathode ray tube, teletypewriter, or an optional programmable, printing calculator. The second connector is the terminus of 16 bi-directional data lines, 8 of which are control lines, giving the user a general purpose programmable interface. This connector also provides power, control, and data signals so that digital data in ASCII FSK may be recorded on and read from an inexpensive audio cassette recorder. All ground- station functions are controlled by a built-in, 8-bit microcomputer (8080A) . The ground station also processes Airsonde signals as described elsewhere in this report. It may be operated on 12 V d.c. or 120 V a.c. (220 V a.c. option available). 20.2.3 Winch The electric winch is enclosed in a suitcase-size aluminum case that protects all parts, including the line, during shipping and storage. When in use, the lid of the case and a small A-frame (which folds inside the case, Fig. 20.2) form the support for a line guide a short distance above a second line guide on a level-wind. The level-wind assures proper winding of the tether line on the winch drum. Controls on the winch are contained in a small hand-held box at the end of a cable. This enables the user to stand under the balloon as shown in Fig. 20.1 while launching and recovering the sensor package. The controls consist of a switch with UP, DOWN, and OFF positions and a knob with which to control the speed of the drum. 20.2.4 Ancillary Equipment 20.2.4.1 Additional sensors Five of the eight sensor channels in the time multiplex format are used for standard meteorological variables. The other three can be used to repeat a variable or for additional sensors. 20.2.4.2 Output devices The most common devices for receiving output from the ground station are a chart recorder, a modified Hewlett-Packard printing calculator, and a tape recorder. The analog output is a voltage. To record it a chart recorder should respond to an input of to 5 V and have a full-scale response time of less than 0.5 s. 119 Digital data may be sent to a teletypewriter or a modified HP-97 printing calcu- lator. The modified HP-97 is capable of receiving data, making calculations, and printing both the original and computed data. Since the calculator is easy to program, the user can readily calculate any quantity permitted by the meteorological data from the ground station and the capacity of the calculator. Potential temperature and mixing ratio are two common- ly calculated quantities. 20.3 SENSOR CHARACTERISTICS 20.3. 1 Temperature-Humidity Temperature and wet-bulb depression are measured by thermistor networks which form the sensing elements of a psychrometer . Each network produces a voltage output that is linear with temperature. The manufacturer of the networks claims +0.15°C accuracy and interchangeability , and a deviation from linearity of +0.16°C. The deviation from lineari- ty is a known function of temperature; consequently it can be removed if desired. Both the dry- and wet-bulb temperature sensors are placed along the axis of a double-walled, cylindrical radiation shield that is pointed into the wind. The temperature element is ahead of the wet-bulb element in the flow, and the two are separated enough to prevent evaporation from the wet-bulb from affecting the dry bulb. A small fan at the rear of the tube aspirates the sensors. Water from a small reservoir flows along a wick and keeps the cover of the wet-bulb sensor wet. The circuitry in the sensor package measures temperature and wet-bulb depression; it does not measure wet-bulb temperature. This assures an accurate wet-bulb depression even when temperature or wet-bulb temperature may be changing rapidly with time. If wet- bulb depression is less than 5°C, the deviation from linearity of the depression never exceeds +0.1°C. With a wet-bulb depression of 10°, the linearity error could reach +0.17°C. The time constant (i.e., the time for the sensor to reach 63% of a step change in temperature) of the dry-bulb sensor is about 5 s in the aspirated psychrometer. The time constant of the wet-bulb sensor is believed to be about 15 s, but this value is not well established . Tests under steady-state conditions have shown the Tethersonde psychrometer to be the equal of any good Assmann psychrometer. 20.3.2 Pressure Pressure is measured by a barometer that is similar to the NCAR digital barometer described by Pike and Bargen (1976). The barometer is also essentially the same as the one described in the companion paper on the Airsonde system. In the Tethersonde system, how- ever, pressure measurements are relative. By using a potentiometer in the sensor package, pressure output is adjusted to read zero at the ground. After launch the output is the difference from surface pressure rather than the absolute pressure. 20.3.3 Wind Speed A three-cup anemometer mounted on top of the sensor package turns a small tachom- eter generator. According to the anemometer manufacturer, the threshold is 0.4 m/s, the distance constant is 2.4 m, and the voltage output is directly proportional to wind speed. Wind tunnel tests of the anemometer on a sensor package indicate that the body of the package does not affect the calibration. The response of a cup anemometer to a wind veloc- ity vector that is not normal to the axis of rotation is such that tilting of the axis in horizontal flow has little effect if the tilt angle is 15° or less. This response is summarized by Moses (1968). The tilt of the vertical axis of the anemometer on the sensor 120 package is difficult to observe, but except for brief spells of turbulence it is generally less than 15°. 20.3.4 Wind Direction The balloon is used as a wind vane. When it is tethered at the nose, fully inflated, and the sensor package is suspended just in front of the tail fins, the balloon heads into the wind. This behavior has been verified by tying a light streamer to the tether line a short distance below the balloon and comparing its orientation to that of the balloon. The sensor package is suspended by two lines, one from each side of the balloon. To assure constant orientation relative to the balloon, a rigid spacer about a meter in length is placed between the suspension lines at a distance of 1 m below the tie points on the balloon. A magnetic compass in the sensor package detects the orientation of the balloon relative to magnetic north. The compass needle is momentarily locked in place on a potentiometer each time direction is sampled, and is free to find magnetic north between samples. To determine true wind direction a compass deviation correction must be made. Wind direction is difficult to determine by another method that can serve as a check on the direction measured by the Tethersonde system. Comparisons made with tower data and careful visual checks suggest that in smooth air the Tethersonde indicates true wind direction to within 5°. 20.4 ELECTRONICS 20.4.1 Sensor Package The TS-1A sensor package circuitry is divided into five major sections: (1) voltage regulator and battery, (2) sensors and conditioning circuitry, (3) analog multi- plexer, (4) voltage controlled oscillators and filters, and (5) FM transmitter (Fig. 20.3). 20.4.1.1 Voltage regulators and battery The sensor package is powered by a 12-V, 500 mA-h, nickel-cadmium rechargeable battery. The battery voltage varies from 13.6 V at full charge to 10.5 V, the recommended low discharge point. Three regulated voltages are provided in the sensor package. These are 7 V + 0.01 V, 3 V + 0.01 V, and 9 V + 0.25 V. The 7-V regulator is the primary reference, and both the 3 V and 9 V refer to this voltage. 20. 4. 1 . 2 Sensors Both temperature and wet-bulb temperature are sensed with bead thermistors. Each bead is a composite of two nonlinear thermistors. The composite, with a precision resistor network, produces a voltage that varies linearly with temperature. The voltage from the temperature sensor is amplified and scaled for direct input to the analog multiplexer. The voltage from the wet-bulb sensor is compared electronically with the temperature sensor voltage, and the difference is scaled and amplified for input to the multiplexer. The anemometer cups drive a small d.c. generator whose voltage is linearly pro- portional to wind speed. This voltage is amplified and scaled for input to the multiplexer. Wind direction is sensed by an electronically actuated magnetic compass as de- scribed in section 20.3.4. The compass consists of a circular potentiometer element, a wiper mounted on a small bar magnet, and a coil surrounding the magnet. The Earth's mag- netic field aligns the bar magnet and wiper. Current is passed through the coil, and the 121 (TnT\ DRY AND WET BULB TEMP pw^) f'i^A) [p^y- T o X LU _l Q. _l D _l LU < CJ 2 4UiJMHZ PRESSURE (ANEROID) \ / w w 2500Hz VCO XMTR WIND SPEED *J w PDMTIM- UOUS CHANNEL O— ► 500Hz VCO WIND DIRECTION ^ t t POWER SUPPLY ' ► +7.00v ► +3 OOv w i J HI LO REFERENC E Figure 20.3. Circuitry of Tethersonde, large magnetic field created causes the wiper to contact the potentiometer element produc- ing a voltage linearly proportional to package orientation. Pressure is sensed by an aneroid barometer. The transducing element is an aner- oid capacitor of special design (Pike and Bargen, 1976). This capacitor is a timing ele- ment for an RC oscillator. A second fixed capacitor is also used as the timing element in the same RC oscillator. The two capacitors are switched in alternately. The ratio of the resultant frequencies is proportional to pressure. The pressure transducer produces a d.c. voltage that varies linearly with pressure. This voltage is compared electronically with the voltage from a "zero" potentiometer, and the difference is scaled for pressure change from the initial setpoint. 20.4.1.3 Analog multiplexer A 10-channel analog multiplexer and associated timing logic generate the Tether- sonde frame format. A frame consists of 20 time-clock periods. The frame starts with three periods of high reference. The eight data channels follow, each one period long and pre- ceded and followed by a one-period low reference. A unique automatic scale expansion circuit works in combination with the multi- plexer logic to give a 4:1 gain in the temperature and pressure sensor outputs. The cir- cuit automatically switches the reference at 25%, 50%, and 75% of full scale to give four times the normal resolution for chart record interpretation. 122 20.4.1.4 Voltage-controlled oscillators (VCO) The Tethersonde transmits data on two audio channels. These channels are fre- quency multiplexed to modulate the RF carrier. Two VCOs are controlled by the voltage on the analog buss. One VCO oscillates at a nominal 500 Hz + 75 Hz and is used to transmit continuous data from any individual sensor. The other VCO oscillates at 2500 Hz + 375 Hz and is used to transmit time multiplex data on the analog bus. It carries all the informa- tion processed by the ground station microcomputer. Output from each VCO is a square wave. The fundamental frequency is extracted from each waveform before frequency multiplexing by two low-pass active filters. 20.4.1.5 FM transmitter A crystal-controlled RF transmitter produces 5 mW at 400-410 MHz depending on crystal selected. The transmitter is referenced to a voltage-controlled crystal oscillator (VXCO) running at a nominal 44 MHz. The ninth harmonic of the VXCO is generated through two stages of frequency conversion. The first tripler runs at about 134 MHz and the second at about 403 MHz. The final stage is a power amplifier at 403 MHz and a buffer for the monopole antenna. 20.4.2 Ground Station The ground station is the Model TS-2A-GS, which is identical to the ground station described in Chapter 19 on the Airsonde system. The discussion of ancillary equip- ment such as the printing calculator and the tape recorder contained in that chapter is also pertinent here. Table 20.1 lists typical calculator printouts. Ground station software permits every n-th data frame to be sent to the calculator whereas every frame is stored by the tape recorder. Thus, for example, a 10-s frame rate may be chosen, and every third frame may be sent to the calculator. It then has 30 s to calculate and print a data frame, time enough to produce a printout like that shown in Table 20.1 (left). Since every frame is recorded on tape, and since on playback the ground station can be programmed to send taped data to the calculator at any desired rate up to 99 s per frame, the calculator can do extensive postflight calculations with every data frame. The real-time calculations are often found to be adequate, but in any event they provide information with which to make operational decisions during flight. Then if greater vertical resolution is needed or if additional calculations are desired, the tape record can be used. Table 20.1. Typical output from the TS-1A-PC printing calculator Calculated printout Variable Printout Time (nun) 9.0505 *»» P (mb) 838.2 *** z (m) 121.5 *** T (°C) 23.2 *** RH (Z) 26.6 *** w (g/kg) 5 . 7 *** True Dir (deg) 112.1 *** U (m/s) 2.7 *** 0(kj 311.7 *** BV (V) 12.3 *** Not-calculated printout- Variable Pi l n t ii lit Elapsed Time (min) 1 . 8 *** Prel (mb) 8.4 *** T (°C) 25 . 2 » ■•» AT (°C) 11.9 *** Mag Dir (deg) 321.6 "•'■" U (m/s) 0.6 *** BV (%) 92.7 *** "'•'Printout without calculations, but with the order of the variables changed from the order in which they are received at the ground station. 123 20.5 REFERENCES Morris, A. L. , D. B. Call, and R. B. McBeth (1975): A small tethered balloon sounding system. Bull. Amer. Meteor. Soc. 56(9) : 964-969 . Moses, H. (1968): Meteorological instruments for use in the atomic energy industry. In Meteorology and Atomic Energy , ch. 6, David H. Slade (ed.), USAEC . Pike, J. M., and D. W. Bargen (1976): The NCAR digital barometer. Bull. Amer. Meteor. Soc. 57(9) .-1106-1111. 124 21. TETHERED BALLOON PROFILER SYSTEM K. Stefanicki Institute of Meteorology and Water Management Division of Aerology Warsaw, Poland 21.1 INTRODUCTION To gather more meteorological data from the boundary layer for the Institute of Meteorology and Water Management, a simple and inexpensive sounding system for heights up to about 1 km was put into operation. Figure 21.1 shows the instrumentation package used in the BLIE. 21.2 COMPONENTS OF THE SYSTEM 21.2.1 Radiosonde Table 21.1 lists the characteristics of U. S . S .R. -made radiosonde A-22-IV with redesigned pressure element, an additional wind sensor, and automatic switch for PTU and W signals . Figure 21.1. Instrumentation package used with tethered balloon in the BLIE. 125 Table 21.1. Radiosonde characteristics ics Paramete rs Characterist Pressure Temperature Humidity Wind speed Sensors Aneroid Bimetal Organic Cup anemometer Measuring 1050 to 700 mb -40 to +40°C 10% to 100% 1.5 to 10 m/s range Resolution 1.5 mb 0.5°C 2% 0.1 m/s Data rate 4/min 4/min 4/min 4/min Coding Morse Pulse/10 s Further information about the radiosonde is presented below: Transmitter frequency: 400 + 5 MHz. Power supply: 4.5-, 9-, and 210-V batteries. Ventilation: Natural if wind speed is >2 m/s, artificial if wind speed is <2 m/s; horizontal ventilation duct. Fastening: Metal circular frame with shock absorbers. Radiation shield: Two shields, metal and fiberboard. Weight: 3.0 kg. 21.2.2 Tethered Balloons and Winch Balloon Delacoste (made in France) has a volume of about 12 m 3 and a height limit of about 1 km and is filled with helium. This balloon may not be used when wind velocity exceeds 4 m/s at the ground and 10 m/s aloft. Balloon Zodiac (made in France) has a volume of about 20 m 3 and a height limit of about 1.5 km and is filled with helium. This balloon may not be used when wind velocity exceeds 4 m/s at the ground and 15 m/s aloft. The winch (our own design), which operates electrically, has two cable speeds, 1 and 2 m/s. It is equipped with a hand brake, a cable length counter, and an azimuth disc. For the BLIE the smaller 7-m 3 balloon and winch described in Chapter 20 were used. 21.2.3 Acquisition and Processing Equipment The equipment consists of a receiver, a pulse counter, and a calculator, whose characteristics are listed below: Power requirements: 220 V, 50(60) Hz. Pressure, temperature, and humidity: Audio-Morse signals, recorded manually. Wind speed: Pulse counter, recorded manually. Processing: With the aid of a calculator, height is computed either with the baro- metric formula or by radar tracking; the height of the lowest levels can be determined by cable length. 126 21.3 SOUNDING PROCEDURE The radiosonde is mounted on a tethered balloon. Measurements are made during balloon stops at selected levels, both during ascent and descent. Before launch and just after sounding completion, a ground check of the radiosonde is required. The measurements at each level take about 3 min. 21 .4 RESULTS The results are presented in the form of computer-generated tables and graphs (parameter vs. height). Time required to process the data is about 1 h. Data accuracies are listed below: Height: +15 m . Temperature: +0.5°C. Humidity: +5%. Wind speed: +0.2 m/s. Wind direction: +20° (rough estimation based on determination of balloon position with respect to the north using a special attachment at the winch). 127 22. BOUNDARY LAYER PACKAGES FOR TETHERED BALLOON M. Hayashi and 0. Yokoyama Nation Research Institute for Pollution and Resources Ukima , Japan 22.1 INTRODUCTION Three types of boundary layer packages for balloon-borne measurements were compared in the BLIE. Figure 22.1 shows the prototype, Meisei Denki , model CBS-W-5 . The other two are variations of the prototype: (1) the simplest version, Makino (Fig. 22.2), and (2) the fully equipped version, Kaijo (Fig. 22.3). Although their appearances differ, the variations in sensors are minor. The packages use the same cup anemometer. Meisei uses a bi-directional vane and detects the angle of the vane by the optical method; the others have a vertical vane and detect the angle by a contactless potentiometer (see Chapter 20). Kaijo has a hot-wire anemometer and a thermocouple to measure fine fluctu- ations in wind speed and temperature. 22.2 SENSORS AND DATA PROCESSING 22.2.1 Cup Anemometer Wind speed is detected by a small, three-cup anemometer. The cup is made of molded plastic. The length of the arm is 6.5 cm, the cross-section of the cup is 2.0 cm 2 , and the estimated moment of inertia is 450 g cm 2 . One rotation produces eight pulses on the photoelectric chopper. The pulses are shaped into square waves and mixed with other signals through subcarrier (voltage-controlled) oscillators. 22.2.2 Vane Vertical and lateral wind direction fluctuations are detected by a bi-directional vane. It has four rectangular flat plates (8x9 cm) made of light wood (balsa) with plastic support. The vertical and lateral angles of the vane are detected by using an optical mechanism for reducting the friction. 22.2.3 Compass and Mean Wind Direction The azimuth of the instrument is detected by a magnetic compass. The angle of a small rotating magnet in an oil box is detected by an optical servomechanism. 22.2.4 Thermometer Temperature is detected by a fine platinum wire of 100 Q. The temperature fluctuation is detected by a thermocouple of copper-constantan in the Kaijo instrument. The standard temperature is given by a transistor thermometer with a time constant of 3 min . 22.2.5 Humidity We do not yet have a suitable humidity sensor. The gauze-covered wet-bulb thermometer yields mean humidity. A lightweight hygrometer with fast response and low electric power consumption is needed. 128 o BI-VANE o A CUP ANEMOMETER 100 i^mf THERMOMETER \ : ■■ i i -■ mm 40 mm VERTICAL VANE -~=sS£ THERMOMETER ooooooooooo p CUP ANEMOMETER Figure 22.1. Prototype boundary layer package for balloon-borne measurements Figure 22.2. The simplest version of the prototype. CUP ANEMOMETER ^Qv ^ iQv VERTICAL 1 VANE BATTERY RECORDER 20mm COMPASS HOT WIRE THERMOMETER THERM0- ) COUPLE Figure 22.3. Fully equipped version of the prototype, 129 22.2.6 Height The height of the instrument package is given by released line length with allowance for elevation angle to the balloon when the wind speed is high. 22.2.7 Data Processing System The electronic data processing system in the package is diagrammed in Fig. 22.4. The signals from each sensor are converted to frequency modulated signals according to the IRIG standard and then mixed together. The mixed FM signal is transmitted by frequency- modulated UHF carrier at 404.5 MHz. However, wireless transmission of this type is per- mitted only for experimental tests. 22.2.8 Recording System The mixed signal is recorded on an audio microcassette tape recorder, Sony M-200, for practical use (Hayashi et al., 1974). To avoid the wow and flutter of the tape re- corder, a standard frequency (3 kHz) is recorded simultaneously. This is used as an au- tomatic frequency controller when the signal is reproduced. The recording duration is 30 min. The recorded signal is reproduced in the laboratory, discriminated by filters, and reduced to the original signal through a frequency-voltage converter. 22. 2 . 9 Harness The instrument package is hung on the mooring ropes below the center of the cap- tive balloon shown in Fig. 22.5. The balloon is tear-drop shaped, with four fins at the rear. The volume of the balloon varies from 20 to 75 m 3 . The mooring rope is nylon fish line 6 mm in diameter, 0.0142 g/m, and 500 kg test. The instrument is supported by a gimbal bar so that all sensors are exposed to the wind. For the BLIE, the 7-m 3 balloon and winch described in Chapter 20 were used. 22.3 CALIBRATION 22.3.1 Cup Anemometer The characteristics of the cup anemometer have been studied in a wind tunnel and in the field by Hayaski and Miyake (1973). The equation of motion of a cup anemometer is T dw ^ _ ,, I - + F = M, where I is the moment of inertia, w the angular velocity, F the friction, and M the torque of the rotating assembly of the cup. The torque is assumed to be given by the difference in the drag of a pair of cups: M = | pRS{C_(U-u)R) - C + (U+U)R) }- , where U is the wind speed, R the length of the arm, p the air density, S the cross sec- tion of the cup, N the number of cups, c_ and c. the drag coefficients of the cup along and against the wind, and a the variability of the torque of the entire cup assembly during a complete revolution. If N = 2, a varies between and 1, but for three- and four-cup ane- mometers a is nearly uniform and equal to or slightly less than 1. The drag of the rota- ting system is divided into static and dynamic terms as follows: 130 CUP ANEMOMETER O-rO BI-VANE COMPASS THERMISTOR 700HZ -* j 73QHz] — » ^>-»J960HzL-* 2300HZ ^ll ECORDER REPRODUCER* E z 2 (X o in Q ANALOG TO DIGITAL CONVERTER 1 TEAC DP -5000) WIND SPEED VERTICAL ANGLE HORIZONTAL ANGLE TEMPERATURE DATA ANALYSER t H I TAC -10 > (-►MEAN VARIANCE SPECTRUM FLUX COSPECTRUM SUB-CARRIER OSCILLATOR FIELD LABORATORY Figure 22.4. Data-processing system for the balloon package, Figure 22.5. The instrument package mounted under a captive balloon. 131 RPS o LU LU CL tO Q_ Z) U 12 3 4 5 6 7 WIND SPEED (m/s) Figure 22.6. Calibration of cup anemometer in wind tunnel. Tunnel wind speeds were measured by a sonic anemometer calibrated by a Pitot tube. F = The fluctuation of wind speed, u, is smaller than the mean wind speed, U: U(t) = U + eu(t) and Rol(t) = V + 8Vj(t) + £ 2 v 2 (t) + where e gives the order of the perturbation. The zero-order equation gives the calibration curve of the cup anemometer. The relationship between the wind speed and the rotation of the cup assembly was tested in the wind tunnel. The Pitot tube was used as the standard instrument for measuring the flow velocity. At lower wind speeds, since the output of the Pitot tube is difficult to read, we used a sonic anemometer, Kaijo model PA-100, as our standard instrument. The result of the test is shown in Fig. 22.6. The starting wind speed is about 10 cm/s. The transient response of the cup anemometer was tested in the wind tunnel. The first-order perturbation gives the linear response for wind speed change. The response is given by a distance constant aNpR 2 SVc_C+ 132 Figure 22.1. Comparison of mean wind speed measured by a cup anemometer and a sonic anemometer in turbulent atmosphere. The plotted letters identify different runs. U cup (m/s) This response is about 4.5 m for this cup anemometer. The second-order perturbation shows the nonlinear response of the cup anemometer, i.e., the over-run (rotation) of the cup assembly. This is one of the most distinctive features of a cup anemometer, i.e. , that it overestimates the mean wind speed in turbulent air flow. The cup anemometer calibrated in the wind tunnel was compared with the sonic anemometer, Kaijo model TR-32, in the real atmosphere. Both instruments were mounted on top of a 13-m tower. The overestimate of the cup anemometer amounts at most to 1%, as shown in Fig. 22.7. The relationship between the wind speed, U, and the cut-off frequency, f , is obtained by comparing spectra. The cut-off frequency is U/2nL. The comparison gives the frequency response function of the cup anemometer, as shown in Fig. 22.8. The turbulent intensity measured by the cup anemometer is about 83% of that measured by the sonic anemom- eter, as shown in Fig. 22.9. 22.3.2 Vane The transient response of the vane was tested by Yokoyama (1969) with a wind tunnel. Figure 22.10 shows an example of the time response of the vane to a step-function direction change. The response of the vane angle, 0, can be represented by the following equation of simple damped oscillation: d 2 „ , d0 dt^ " 2 ^ dT , 2 = F(t) where u n is the natural angular frequency of the vane, £ is the damping ratio (the ratio of the actual damping to the critical damping), t is the time, and F(t) is a time-dependent forcing function. The damping ratio, t,, and natural frequency, w are obtained from the measurements of the time response to a step-function direction change for various wind speeds by use of a solution of the equation for the case of t, < 1. The distance constant of the vane, U/oj n ^, is used as an index of the response characteristics and regarded as a constant for high wind speeds. It is approximately 1.6 m for this type of vane. An example of the frequency response for typical wind speeds is shown in Fig. 22.11. 133 10 G 2 (f) ' 0.1 001 0.001 4 17 001 C I A %s *X f /fc 10 Figure 22.8. The frequency-response function as a ratio of power spectra from the cup and those of the sonic anemometer. The plotted letters and numbers identify different runs. SAT Figure 22.9. Comparison of a u /U as measured by cup and by sonic anem- ometer. The plotted letters iden- tify different runs. CUP 134 Time Figure 22.10. Example of the measured time response of the vane to a step- function direction change. (Mean wind speed is 3.0 m/s and angle of attack is 45.0°. ) o 2 1 5 2 o II I \ \ 1 1 \ 5.0 \ N^ 50 - 0. l\ \l 1 >S S > v I I I I I I II I I I II I II I 1 0.5 1 90 -180 a -9- 10 n (Hz) Figure 22.11. Frequency response of the vane for various wind speeds. (G is the ratio of amplitudes and - Dry Wet T Press Zero T I y&. 1 II It II II II a Aneroid Wind Sp J*^ Wind Dir. Down Level Ground Station V Receiver —I— Audio Discriminator Strip Chart Recorder Cassette System Filter Cassette | counter I frequency Recorder | I Detector Real-Time I Clock I Microcomputer System With Graphics Figure 23.1. Block diagram of NCAR Boundary Profiler electronics. of each data frame time is recorded on the magnetic tape. Figure 23.1 is a block diagram of the BP system. 23.3 SENSORS Bead thermistors are used to measure dry- and wet-bulb temperatures. The sensors are mounted in a radiation shield tube and are aspirated by a small fan located at the end of the tube. The wet-bulb thermistor is covered by a sock, which is connected to a small reservoir . Pressure is measured with an aneroid capsule. The pressure sensor is set up to make a relative reading, not an absolute one. The relative reading has a total differen- tial range of 100 mb . Wind speed is measured with a three-cup anemometer connected to a small d.c. generator. Wind direction is determined by using the balloon as a wind vane in conjunction 137 Table 23.1. Sensor characteristics Sensor Total range Set range Precision Thermistor Aneroid capsule Cup anemometer Magnetic compass and balloon -30° to 45°D'- Relativet 0.5 to 10 m/s 0° to 360° 25°C 100 mb to 10 m/s 0° to 360° +0.5°C +1 mb +0.25 m/s +5° "'•'Six switches set temperature ranges in 25°C increments. tPressure is set at ground level relative to launch site. 100% Sync (3.4 kHz) Wind dir Dry T „ _ Dry T Press Wind speed K^ Wet T Wet T 0% (2.6 kHz) 2 !erc ) Down (2.3 kHz) U Figure 23.2. Data frame on strip chart recorder. with a magnetic compass. The magnetic compass is a potentiometer with the compass needle acting as the arm of the pot. When it is time to sample wind direction an electric field locks the needle to the pot windings. This gives a clean reading when a sample is taken. Table 23.1 gives more information about each sensor. 23.4 DATA FORMAT AND PROCESSING Figure 23.2 is an example of one data frame recorded on the strip chart. The average frame duration is 30 s but can range from 20 to 40 s. The sync pulse represents 100% or 3.4 kHz, whereas the zero pulse is 0% or 2.6 kHz. Data are transmitted in the following order: (1) sync, (2) dry-bulb temperature, (3) wet-bulb temperature, (4) pres- sure, (5) zero, (6) dry bulb, (7) wet bulb, (8) wind speed, and (9) wind direction. The user can change this sampling order. The down signal is used to separate each channel sample from the next one. The scale ranges for each data parameter are listed in Table 23.1. The zero reference level for temperatures can be set by the user in 10°C increments 138 BA0/UM0 AUGUST 30,1979 FLIGHT (10) TIME DBULB UBULB DP PRESS US un RH HEIGHT HT/MSL (H:H:S) CM o o o o o o O MO CM o o o u~t i-h o o o CM CM CM CT\ LTl i-H OHM o o o o a ■a OJ OJ M a; (0 ,M a ■i-t a, L. c, -H T3 3 lj (1) OHM Cu C ■a to OJ »£ a, c u o u u tO 0) r-H a nj w to e to OJ to 3 J3 O O C O -H 01 i-H !-i X! r-t 41 MA Xi ■O 4-1 D'O d ■H 11 4- i-4 £ 'H I tl ■OJS l< O 4-1 3 E OJ to 4-1 10 to 0) •t- <_> to xi u 05 Z to 00 -H PL| J3 -H "O 1 — 1 PL, i-i d z PQ t0 *4H -r-t J3 O ! CM 00 H 5= 1 146 ■ fEM Figure 24.5. Vaisala cup and vane system at the 22-m level. examine the effect of separation to determine if the 20-min averaging specified for BLIE is effective in reducing this variability. Terrain irregularities of length scales comparable to separation distances between sensors can introduce systematic differences between their measurements. These differences cannot be removed by time averaging. The terrain around the BAO site rolls gently down toward the west and north but is relatively smooth. Thus the effect of the slope should be very small. Obviously the sampling volumes of the different systems are not at the same distance from the tower. The bistatic systems have their sampling volumes oriented verti- cally over the central transmitter (or receiver), but the tilted beams of the monostatic systems point away from the tower toward the west and north, so their separation from the tower increases as a function of height. 147 1 4 i i r Time ^^W' v/V ^V I L I O Unstable, 1 < u < 5 m/s D Unstable. 5 < u < 15 m/s A Near neutral. 1 < u < 15 m/s Ut+At Note: Points include data from 10- and 300-m levels. -fi. 10 12 T/At ( - 7-Q/d) Figure 24.6. Simulated two-point variance for a downstream separation of sensors plotted as a function of averaging time. Plotted here are variance ratios for t = 1/6, 1/2, 1, 2, 6, and 10 min with At fixed at 1 min. The effect of averaging can be demonstrated for the simple case where the two sensors are aligned along the direction of the mean wind. If the flow is assumed to be horizontally homogeneous and the distance between the sensors small enough that the wind field can be considered frozen, the spatial separation converts to a time lag At = d/u, where d is the distance and u is the mean wind. For typical d = 300 m and u = 5 m/s, the lag corresponds to 1 min. In a truly frozen field, 19 of the 20 min would be the same in both measurements, bringing the two averages very close together. The above assumption is tested in Fig. 24.6 with single-point time series of the stream-wise wind component, u, measured at two levels on the BA0 tower. The variance of the difference across a fixed time lag, normalized by -e.se ■ 1 .00 -1 00 -0'.5@ oe el 5c-i i . V-COMPOMEUT O i — I TO > ei .so_ e.ee_ i i / e.se_ 1 .00 - _ I o o -i.ee -0.50 9.00 0.50 i .en BAO ( x 10) RELBTIVE HUMIDITY I -i 1 1 r X 0.80 1. 20 0.48 8.6C BAO o X Restricted Wind Directions 1 .00 1 .56 TEMPERHTUPE 2 .00 BAO £.50 3 '. 8 (x 10) o X TO > All Wind Directions 1 .50 2.00 2.50 BAO (x 10) o 1 1 1 / X w 1 .58_ / 1 .00. TO TO 0.5@ •H TO > : $jH&'' - 0.00_ -0.S0 1 1 1 1 (x 100) -0.50 0.00 0.50 1.00 1.50 2.0e BAO (x 10) Figure 25.2. Scatter plots of data from Vaisala tower sensors vs. data from BAO tower sensors. 161 Notes on Fig. 25.3 Comments from P. Martin, EERM 1. SAM-B temperature readings during flight appear to be lower than tower temperature readings by about . 4°C for ascents and 0.2°C for descents. Although this may be caused by calibration error, the probability is low. 2. In a 20-min period when temperatures are rising, instantaneous temperatures ob- tained from SAM-B at the beginning of the period will be lower than the mean temperature for that period. 3. Consideration must be given to possible perturbations of the temperature field caused by the tower or the lack of ventilation in tower sensors. Editors' response The temperature sensors on the tower are aspirated and shielded. Periodic comparisons between measurements on the tower and the NCAR aircraft show agreement at 250-m height to within ±0.1°C. Radiation error in the tower measurements is less that 0.04°C. Notes on Fig. 25.4 Comments from D. Martin, EERM Temperature readings from SAM-B instrument package on carriage are higher than tower temperatures because of lack of adequate ventilation in our sensors while on the carriage. This difference in behavior proves the good ventilation on the SAM-B aircraft. 162 TEMPERATURE TEMPERATURE O CO 1.515. Aircraft Ascent •V 1 .00 1 .5G DEU POINT BAO (x 10) o X 1 .50_ Aircraft Ascent -0.50 0.00 0.50 1.00 1.50 2.00 BAO (x 10) RELATIVE HUMIDITY O o x a.ee Aircraft Ascent 0.00 0.20 0.40 0.60 0.80 1.06 BAO ( x 100) o X Aircraft Descent 1 .50 2.00 2 .50 BAO (x 10) DEU POINT O X 1 1 1 Aircraft 1 y 1 .58. Descent / 1 .00. t :i~.- 0.50j / *** - n.0O_ /r 6.50 , 1 1„ -.1 __ . 1 - 1 . BAO (x 10) RELATIVE HUMIDITY o o X 0.80_ Aircraft Descent .00 0.20 0.40 BAO .60 0.80 1 .30 (x 100) Figure 25.3. Scatter plot of data from SAM-B air-borne sensors vs. data from BAO tower sensors. 163 TEMPERATURE TEMPERATURE O Carriage 1 1 Ascent y\ 2.50. ■/ - 2.68. / 1 .50. - 1 .60 1 .00 1 '.50 2 '.00 2 '.50 3. BAO (x 10) D F U POINT o y. 1 .50. CQ Carriage Ascent -0.50 0.00 ).5B 1 .00 1 .50 2.00 BAO (x 10) RELATIVE HUMIDITY o o Carriage Ascent CQ 0.80 1.00 (x 100) m 1 .511 1 .00 1 .50 2 .00 BAO DEU POINT O X 1 .50. Carriage Descent ).S0 0.00 0.50 i .©e BAO RELATIVE HUMIDITY O O pq I 3.00 (x 10) (x 10) 1 1 Carriage i i 0.80_ Descent - 0.e0_ 7"! / /■ ••'* -■ 0.40_ 0.20 0.00 / .00 B 1 . 20 0'.40 e',68 e'.sa i . Figure 25.4. Scatter plots of data from SAM-B instrument package on carriage vs. data from BAO tower sensors. 164 U-COMPONENT U-COMPONENT o I— I 0.00_ i-H -8.59. -1 .80., -1.88 -0.58 8.80 0.50 1.88 BAO (x 10) V-C0MPONENT API T O X r-, " 0a < -0.se_ -8.50 0.00 BAO 0.50 1.00 (x 10) O X < 1-1 < 1 0.50. .^ ' ' e.eg. / €3> 2 e.58_ - 1 .88 - - - 1 _ - I -1.80 -0.58 0.00 0.50 1.88 BAO (x 10) V-COMPONENT l T O i— i X 3 < 0.00. -0.5Q. -1 .08.^ 0.50 1.08 (x 10) Figure 25.5. Scatter plots of data from TALA 1 and TALA 2 vs. data from BAO tower sensors. Tala 1 is the small kite flown during light winds, and TALA 2 the large kite flown during strong winds. Notes on Fig. 25.5 Comments from C. F. Woodhouse, Approach Fish, Inc. 1. The data points in group 1 should be discounted since they were obtained from an uncalibrated lift balloon flown at 800-m altitude. The velocity points shown are only empirical estimates. 2. The data points in group 2 should be discounted because of equipment malfunction caused by an attempt to protect the electrical contacts with wet silicone in a thunderstorm. 165 U-COMPONENT TEMPERPTURE o S i i i e.se . * */ • e.ee • - ■ , * e.59. / /• .V - 1 .00 i i -i.ee -0.58 e.ee e.se 1.00 BAO (x 10) V-COMPONENT 1 1 1 e.se_ • y " e.ee_ • / t • - 0.50. / • • - 1 -00_ / • -1 .00 -e'.se e'.ae 0'.50 1 o CO w 1.50_ 1 .94,, BAO (x 10) Figure 25.6. Scatter plots of data from FM-CW Doppler radar and RACES vs. data from BAO tower sensors. Notes on Fig. 25.6 More data were collected with the FM-CW radar pointing west than pointing south, which accounts for the disparity in the number of data points along the two components. 166 U-COMPONENT U-C0f1P0NENT O X 0.04 50 1 .60 (x 10) V-COMPONENT 1 1 / t .50_ t • : 'Mi- ' ** 0.00 ** /W* W f , JA? t <. !>,• -a. 69, /' - -1 .00, / 1 -i.ee -e.5e 0.00 0.50 1.00 BAO (x 10) o r— I X ^. i.se 0.00 0.50 1 .00 BAO (x 10) V-COMPONENT O > < Figure 25.7. Scatter plots of data from WPL and AVIT Doppler sodars vs. data from BAO tower sensors. Notes on Fig. 25.7 1. WPL and AVIT data were possibly affected by crosstalk when the sensors were operated concurrently for the first 3 days. 2. The 50-m WPL wind readings were raised by a factor of 2 to correct for attenuation caused by array geometry. 3. During high winds WPL Doppler measurements were contaminated by tower noise (see discussion in Chapter 11). 4. Vertical velocity correction (see section 12.3) was not applied in the wind computations of either system. For 20-min averaged U and V, this correction provided no detectable improvement. 167 U-COtlPONENT ll-COMPONEHT -1 .00 t -1 .8 -0.50 0.00 BAO V-COtlPONENT 0.50 1.00 (x 10) o X -o G o w o -d -0.S0 u UJ •1 .00_; o X 3 -C UJ 100, 150, and 200 m , i ■ .. <** » y V / / / / rr. -r -1.00 "0.50 0.00 BAO V-C0MP0NENT 0.50 1 .00 (x 10) i.se_ (x 10) Figure 25.8. Scatter plots of data from Echosonde Doppler sodar vs. data from BAO tower sensors. Data from 100, 150, and 200 m show better agree- ment with BAO tower sensor data than those from levels above and below. Notes on Fig. 25.8 Comments from M. McAnnally, Radian Corporation 1. The low bias in the Echosonde wind data is caused by an improper transmit antenna pattern resulting from an incorrect horn shape and a phasing mismatch between transducer driver pairs used with each horn. Combination of these two factors resulted in a transmit beam pattern that was much broader in azimuth than anticipated. As a result only the alti- tude range from 100 to 230 m received adequate acoustic energy. 2. New transmitter designs have been incorporated into the system, and subsequent testing has provided excellent agreement with observed conditions. 168 U-COMPOHENT U-CuMPOHENT c X All Levels -e.sa_ 10 -0.50 0.00 BAO -COMPONENT ) .50 1 .00 (x 10) BAO .50 1 .00 (x 10) Q .06. en 100 and 150 m -1 .00 -0 .50 V-COMPONENT BAO o 0.50 1.0G1 (x 10) (x 10) Figure 25.9. Scatter plots of data from Sensitron Doppler sodar vs. data from BAO tower sensors. Data from 100 and 150 m show much better agreement than those from levels above and below. Notes on Fig. 25.9 Comments from S. Salomonsson, University of Uppsala 1. The large scatter in the data results from using borrowed (WPL) antennas which we could not tilt more than 22° from the vertical. The measurements may be influenced by the vertical wind component, since correction for that effect was not applied. 2. Our low pulse power (100 W) resulted in weak echo signals above 150 m, whereas a ringing in the transducer affected measurements below 100 m. Only measurements at 100 and 150 m are therefore considered appropriate for comparison with tower data. 3. Our preliminary system is not optimized for sensing at high altitudes but rather to explore the PLL technique in Doppler applications. Agreement with tower at 100 and 150 m indicates that the PLL circuit works fairly well when the echo signals are strong. 169 U-COMPOHENT V-COMPONENT c X c X I A 0.50_ • ■ # - ■ 0.00 m »%■•> . •w ;&? '£/ "' 0.50. «• •/ 1 .00 -l 00 «'. '.0 .00 0.50 i . o o X 1 i 0.56. • .V ' 0.00 • **t^- •' • :.f- 0.50_ /. - 1 .00 -1 .00 -e'.sa 00 «'. 50 1 . BAO (x 10) BAO (x 10) Figure 25.10. Scatter plots of data from XONDAR Doppler sodar vs. data from BAO tower sensors. • 50-m level data; ■ data contaminated by severe tower noise . Notes on Fig. 25. 10 Comments from R. L. Peace, Jr., Xonics, Inc. 1. The plot contains data points known to be affected by tower-generated wind noise and possible crosstalk from two acoustic sounders not officially participating in BLIE. We at Xonics would like the data flagged and identified. 2. Use of 160-ms-long transmit pulse (rather than the 80-ms-long pulse used in a few soundings) resulted in a strong bias of the 50-m-level data toward zero velocity. This problem was not recognized until late in the first week. 3. We would like to point out discrepancies of the order of 0.5 to 1 m/s in the wind speed and 10° in wind direction between the BAO sonic anemometer and Propvane when the wind direction was between 20° and 120° and between 180° and 300°. Editors' response The unsatisfactory 50-m data and those affected by wind noise are easily identified and flagged, but points corresponding to the two periods with possible crosstalk are too close to the 1-to-l line to benefit from flagging. Discrepancies between measurements made from opposite sides of a tower are not surprising (see discussions in section 25.1). To minimize the effect of flow distortion caused by the tower, only measurements from the upwind boom were used in the plots. Also, note that computer summaries of Propvane wind components made available to all participants were computed from the 20-min averaged speed and direction rather than from the instaneous wind components. Such averaging introduced discrepancies between the sonic anemometer and Propvane statistics for low wind speeds (when turbulence intensity is high). However, the Propvane data used in the comparisons described here are true components recomputed from the raw data. 170 TEMPERATURE TEMPERATURE o X N H > I 6 .flO 1 .50 2 .00 ; .56 "5 .00 BAO (x 10) DEU POINT O i — I X 10 to 300 m yU&'r ft y -r -0 .58 0.60 0.50 1 .1 BAO RELATIVE HUMIDITY 1 .50 2 .00 (x 10) o c X 9.8B_ > 9. C3 10 to 300 m 0.29 0.4O 0.60 0.80 1.00 BAO (x 100) o X! H > i .oe_t 00 1 .50 2.00 2.50 3.00 Sonde Average (x 10) DEU POINT o X N H > e.5e O 0.00 10 m to 3 km j* 1 if . -0.5O e'.oe 0.50 r.00 r.50 Sonde Average (x 10) RELATIVE HUMIDITY o c X 0.8< IS] H > 0.40 I a O .20 10 m to 3 km 0.00 0.20 0.40 0.60 0.80 1.00 Sonde Average (x 100) Figure 25.11. Scatter plots of balloon-borne GMD-1/VIZ radiosonde data vs. BAO tower data and vs. average for all radiosondes. Data in the righthand group are restricted to periods when all four sondes were functioning sat- isfactorily. 171 Notes on Fig. 25 . 11 Comments from R. B. McBeth, NCAR 1. Points in group 1 are too high because the operator failed to apply the frequency drift correction. 2. Points in group 2 are 10-m temperatures at about 2315 MDT . The sondes in these two instances were launched before they had come to thermal equilibrium with the cold outside air. Measurements from both sondes agreed within 0.4°C with the tower data during the carriage ascent that preceded the launch. 172 TEMPERBTURE RELQTIVE HUMIDITY O X 2.09. I .50. 1'.50 2.00 2.50 3.00 BAO (X 10) DEU POINT X —I I Q 1 .50 2.00 (x 10) o o 0.80_ H > O - 2 C c o H > . 00_ t 0.00 0.2 BAO (x 100) 1 Free Ascent 1 / 10m to 3 km f / 2.00 / / / ' 1 .00. / / / r .00 ^ , i 0.00 I .00 Sonde Average :.00 3.00 (x 1000) Figure 25.12. (Upper, left and right, and lower, left) Scatter plots of data from GMD-1/VIZ radiosondes on carriage vs. data from BAO tower sensors. (Lower, right) GMD-1/VIZ radiosonde computed heights vs. average heights for all balloon-borne radiosondes. Notes on Fig. 25.12 Comments from R. B. McBeth, NCAR The very low values in group 1 are believed to result from computational error. When the same sonde was flown from the balloon, its measurements averaged 0.13°C higher than the average data of the other three sondes and 0.19°C lower than those from the tower. 173 TEMPERATURE TEMPERATURE C X in n H 10 1 to 1 300 m 1 ^>*' 2.50_ - 2.O0_ x** • 1 .S8 - i .00 /^ i .08 l'.50 2 '.00 2.50 3. BAO (x 10) DEUI POINT O — -0.E0 0.00 0.50 1.00 1.50 2.30 BAO (x 10) RELATIVE HUMIDITY 0.90 0.20 0.40 BAO .60 6.30 1 .00 (x 100) o X CO O H 1.00 1.58 2.00 2.58 3.00 Sonde Average (x 10) DEU POINT o X 1 1 10 m to 3 km i 1 W 1 .S0_ - 1 .00. * ir 1 _ 00 P 0.50 0.00_ „-' >.7& " ' i - -0.50 A' -0 SB e'.ae e'.se 1 '.00 1 .50 2. Sonde Average RELATIVE HUMIDITY T (x 10) 0.20 0.40 0.60 Sonde Average 0.80 1.06 (x 100) Figure 25.13. Scatter plots of balloon-borne TDFS radiosonde data vs. BAO tower data and vs. average for all radiosondes. Data in righthand group are restricted to periods when all four sondes were functioning satis- factorily. 174 Notes on Fig. 25.13 Comments from E. Schollmann, Deutscher Wetterdienst 1. The TDFS data have no radiation error because the sensors are well aspirated and housed in a radiation shield. The sensors are bead thermistors with fast response. 2. Discrepancies between the sonde and the tower profiles may result because data are not compared at significant points but at fixed pressure steps. TEMPERATURE RELQTIVE HUMIDITY O i — i Q H 1.50 2.00 BAO 2.50 3.00 (x 10) DEU POINT o X CO Q H -0.50 0.00 0.50 1.00 1.50 2.00 BAO (x 10) 0.20 0.40 0.60 0'. BAO (x 100) o o o H Free ascent 10 m to 3 km 0.00 1 .00 2.00 3.00 Sonde Average (x 1000) Figure 25.14. (Upper, left and right, and lower, left) Scatter plots of data from TDFS radiosondes on carriage vs. data from BAO tower sensors. (Lower, right) TDFS radiosonde computed heights vs. average heights for all balloon- borne radiosondes. 175 TEMPERBTURE TEMPERQTURE O X 2 c 1 i 10 to 300 m s.sa. jf- * 2.38. / , , ' - 1 . 53_ '• * - 1 .00 i i ' 1.03 1.50 DEU POINT BAO o 10 to 300 m 2.50 3.E (x 10) i r .50 0.00 0.50 1 .00 1 .E BAO (x 10) RELATIVE HUMIDITY 0.80 1.03 (X 100) o I— I X 2.00_ o 1 .00.; .00 1.50 2.00 2.50 3.00 Sonde Average (x 10) DEU POINT t 1 r 10 m to 3 km J. 50 0.00 0.50 1.30 1.50 2.00 Sonde Average (x 10) RELATIVE HUMIDITY .00 0.20 0.40 0.60 Sonde Average 0.80 1 .00 (x 100) Figure 25.15. Scatter plots of balloon-borne CORA radiosonde data vs. data from BAO tower data and vs. average for all radiosondes. Data in the righthand group are restricted to periods when all four sondes were func- tioning satisfactorily. 176 Notes on Fig. 25.15 The CORA system measured relative humidity with a Humicap sensor that has a faster response than the wet-bulb systems used in the TDFS and Airsonde. TEMPERATURE RELQTIVE HUMIDITY O O u 2.83. 1.5B. (X 10) DEU POINT O X i .se_ O u -e.se. , 8.40 B.68 BAO 8.88 (X 100) o o o o u Free i Ascent 1 10 id to 3 km 2 .00_ 1 .80_ / 0.80 8 .08 l'.00 2 '.00 3 Figure 25.16. (Upper, left and right, and lower, left) Scatter plots of data from CORA radiosondes on carriage vs. data from BAO tower sensors. (Lower, right) CORA radiosonde computed heights vs. average heights for all balloon- borne radiosondes. 177 TEMPERATURE TEMPERATURE o r— I X ..- v 2.50_ 2.80. 1 .S0_ -/'" - 1 .08 i 1 1.00 1.50 2.00 2.50 2.00 Sonde Average (x 10) DEU PGIHT u TJ c o CO !~ ■H < i .se_ -8.58 0.08 0.58 1.00 1.50 2.00 Sonde Average (x 10) RELATIVE HUMIDITY o 1 1 1 1 / o 10 m to 3 km / ■*" . 86. %,/ - e.ie. .V . C 0.40 O S-l ■H •W' < .20 $F - 0.00 (x 100) 8.28 0.40 0.60 0.80 1,06 Sonde Average (x 100) Figure 25.17. Scatter plots of balloon-borne Airsonde data vs. BAO tower data and vs. average for all radiosondes. Data in the righthand group are restricted to periods when all four sondes were functioning satisfactorily. 178 TEMPERDTURE RELATIVE HUMIDITY O c o .04. o o X e.80_ CI C e.4e o w n r Carriage 1 .ee 1 .50 2.00 2 .se BAO (x 10) DEU POINT BAO (x 100) o 1— 1 X 1 .50_ Cj a o -0.S0_ o o c Oj ■o d o VI '-OS 1.50 2.00 (x 10) Free Ascent 10 m to 3 km 1 .00 2 .00 Sonde Average (x 1000) Figure 25.18. (Upper, left and right, and lower, left) Scatter plots of data from Airsondes on carriage vs. data from BAO tower sensors. (Lower, right) Airsondes computed heights vs. average heights for all balloon-borne radiosondes . Notes on Figs. 25.17 and 25.18 Comments from D. B. Call, AIR, Inc. 1. At first, Airsonde temperatures and dew points were often 2°C or 3°C too high in full sunlight but generally accurate at dusk and after sunset. The problem, identified as radiative heating, was corrected by painting the thermistor beads white and the insides of the two ducts black (reverse of original colors). Data from modified Airsondes for 4 and 5 September show excellent agreement with tower data (see Fig. 25.19). 2. The consistent positive bias of the dew points (Fig. 25.17) is caused by absence of any aspiration of the wet thermistor until the instant of launch when the Airsonde began to rotate. The error rapidly diminishes during the first minute of flight and could be eliminated by forced aspiration of the wet thermistor while the Airsonde is on the ground. In the carriage ascents (Fig 25.18) the thermistors were aspirated before data collection, which brought the dew point readings to equilibrium. 179 Notes on Fig. 25.19 Comments from P. B. Call, AIR, Inc. 1. Results are restricted to data obtained with Airsondes modified to reduce radiative heating . 2. On these two days, for the carriage ascents, the wet-bulb thermistor was aspirated by a small electric fan throughout the ascent, beginning a few minutes before ascent. Thus the wet-bulb temperatures are properly depressed, and the dew point data are in good agreement with those of the tower. 3. For balloon ascents, there is little aspiration of the wet thermistor until the Airsonde begins to rotate. Since the wet thermistor requires a minute or two to cool to its proper value, the first few indicated dew points (and relative humidity) are higher than those of the tower. 180 TEMPERATURE TEMPERATURE O X ■a c o Free Ascent 4,5 Sept 1 .50 2 .00 BAO DEM POINT o — X — o < Free Ascent 4,5 Sept o o 0.86. ■c a o Free Ascent 4,5 Sept :.S0 3.06 (x 10) -0.60 0.00 0.50 1.00 1.50 BAO (x 10) PELATIVE HUMIDITY 0.00 0.20 0.40 C.t BAO 0.30 1 .00 (x 100) o X X) c o Carriage 4,5 Sept o x 01 •v a o -0.50 0.00 8.50 1.00 1.50 2.C BAO (X 10) RELATIVE HUMIDITY (.00 0.2 40 0.60 0.80 BAO (x 100) Figure 25.19. Scatter plots of data from Airsondes vs. data from BAO tower sensors for 4 and 5 September 1979. 181 Notes on Fig. 25.20 Comments from A. L. Morris, Ambient Analysis 1. The three plots on the left (tethered balloon ascents) show a positive bias that is absent in the plots on the right (carriage ascents). Since the balloon ascents were made at midmorning and midafternoon whereas the carriage ascents were in the evening, solar radia- tion undoubtedly contributed to the bias. 2. After the intercomparison experiment, the inside of the aspirated duct was painted black to reduce radiation errors. 182 TEMPERATURE TEMPERATURE O X x> a o to S-l Li X) 4-1 id H i l r r Te th ered £.50. J+* {*•' 2.00 >•' "••, - 1 .S0_ *r^ y i / - 1 .00 1 .00 l'.50 2 00 2 '.50 3 DELI POINT O r-H 01 e o to -C D H 1 Tethere 1 1 d i 1 .S0_ * «* ♦•** / - 1 .00^ •:* > * / * * f 1 ' * - 0.50_ * t >'/ - 0.00 * / - a .se_ -0 50 0.00 o'.s© l'.ee 1 .50 2. BAO (x 10) RELATIVE HUMIDITY O o aj T3 o CO OJ xi +J 0.20 CD Tethered 1.50 2.00 (x 10) RELATIVE HUMIDITY O O 0.80. 0.60_ "1 r Carriage BAO (x 100) BAO 3.80 1.1 (x 100) Figure 25.20. Scatter plots of data from Tethersonde vs. data from BAO tower sensors . 183 TEMPERQTURE TEMPERRTURE O X 1 .80 1 .5 DEU POINT BAO 2.50 3.90 (x 10) o I— I 1 .5S_ o c * 0.80 1.50 2.06 (x 10) 0.00 0.20 0.40 0.60 0.80 l.C BAO (x 100) o o 2 0.56. .50 3.00 (x 10) -e.S0_ t -0.50 0.00 1.50 1 .00 1 .50 2.00 BAO (x 10) RELATIVE HUMIDIT'r O c 0.00 0.20 0.4 .60 0.80 1 .06 BAO (x 100) Figure 25.21. Scatter plots of data from IMWM tethered profiler vs. data from BAO tower sensors. 184 TEMPERATURE TEMPERATURE O X a. 1 1 Tethered i •x 2 .50 f/ 2.89. "v 1 .50_ 1 .00 i i i 1.00 1.50 2.00 2.50 3.00 BAO (x 10) DEU POINT o i i i i 1—1 ' Tethered w 1 .E0_ ■ 1 .00. ' y e.59 2; • 1* *** * / * - / a, PQ . ••/ 0.00 ' / -e.58 i i i . i 0.50 0.00 0.50 1.00 1.50 2. BAO (x 10) RELATIVE HUMIDITY O o X 0.80 04 CO o u 1.00 1.50 2.00 2.50 3.00 BAO (x 10) DEU POINT O o -0.50 0.00 0.50 1.00 1.50 2 BAO (x 10) RELATIVE HUMIDITY >0 1 1 1 Carriage BAO (x 100) BAO (x 100) Figure 25.22. Scatter plots of data from BP/NCAR sensors vs. data from BAO tower sensors. 185 Notes on Fig. 25.22 Comments from R. B. HcBeth, NCAR Improper adjustment of the wet-bulb range contributed to the very large dew point scatter. At times the range was not adjusted downward promptly to accommodate sudden drops in humidity. The wet bulb was then reported equal to the lower range limit rather than the correct colder value and the computed dew point read much too high. This does not account for all the scatter; the transfer of water to the wet thermistor was probably inadequate . 186 HORIZONTAL WIND 1 T HORIZONTAL UINTJ BAO (x 10) HORIZONTAL UIND BAO .80 1 .00 1 .20 (x 10) HORIZONTAL UIND O X >.80_ C e.«e o CO !-* tu jS *-> H 0.2Q_ 8.00^^ Carriage 0.00 0.20 0.40 0.60 0.80 1.00 BAO (x 10) HORIZONTAL UIND 40 0.60 .86 1 .1 BAO (x 10) Figure 25.23. Scatter plots of horizontal wind speeds from tethered balloon sensors (balloon-borne and on carriage) vs. data from BAO tower sensors. 187 o o C o W U V X. u QJ H O o c o i .ag. Carriage BAO 2.ee 3.00 (x 100) Figure 25.24. Scatter plots of heights from tethered balloon systems vs. BAO instrumentation heights. 188 25.4 CONCLUDING REMARKS The results of the BLIE presented in this chapter show generally good agreement between the tested sensors and BAO sensors. The amount of agreement differs for each sensor, but differences are usually small. Where strong disagreement was found, the participant provided reasonable explanations for departures and outlined correction pro- cedures. In one instance a participant modified his sensor midway through the experiment and got better agreement for the remaining runs (see Fig. 25.19). The objective of this report is to present BLIE data without interpretation. In evaluating the sensors, the reader is urged to give due weight to the comments and notes accompanying the figures. Hardware and software changes made by a manufacturer after the experiment also need to be considered, since they may have solved problems experienced during BLIE. The success of any experiment involving many different instruments depends on the cooperation between participants and experiment organizers. The authors note with gratitude the grace and generosity with which that cooperation was extended, especially in the preparation of this last chapter of the report. 189 * U.S. Government Printing Office: 1 98 0-O-67 7 -13 8/6 PENN STATE UNIVERSITY LIBRARIES AD0Q072A 32TTA