You are on page 1of 8

Meteorol. Appl.

5, 149–156 (1997)

Design and performance of an acoustic antenna


for a phased array Doppler sodar
Yoshiki Ito, Kaijo Corporation, Sakae-cho, Hamura-shi, Tokyo 205, Japan

An acoustic array antenna which has a narrow main lobe and suppressed sidelobes is studied. The
directivity function formed by the multiple point sources is used to estimate the aperture of an array and
the spacing between the transducer elements to make a desired beam. Then the aperture tapering based
on a Chebyshev distribution and the effect of an acoustic shield are evaluated to suppress the sidelobe
radiation. The total performance of the phased array system has been examined by comparison with
tower instruments. The field test shows that the measured beam pattern agrees with the calculated one
and the sodar-derived winds have a correlation of 0.96 and a standard deviation of differences of about
0.47–0.52 m s–1 compared with in situ measurements. These statistical values are as good as or even
better than those from the previous experiment.

1. Introduction Doppler sodar showed a systematic bias in mean wind


profiles (Vogt et al., 1994) and there was the problem
Measurements of the total horizontal wind in the lower of acoustic emission into the surroundings (Vogt &
planetary boundary layer by acoustic remote sensing Thomas, 1994). Thus, the phased array system has
were first achieved by Beran & Clifford (1972). some room for improvement, especially in its acoustic
Supported by numerous field tests and comparisons performance. It is important to study the conditions
since then, a conventional tri-monostatic or bistatic required to make the beam of the phased array system,
Doppler sodar has been widely used to measure wind because the beam synthesis of the system is highly
profiles up to several hundred metres. A phased array dependent on the combination of acoustic transducer
Doppler sodar, however, has been used often in recent elements, and is more complicated than that of the
work (e.g. Vogt & Thomas, 1994). The author devel- old-type sodar using a parabolic reflector. Based upon
oped the phased array Doppler sodar, which uses five knowledge of the array antennas for radar systems,
beams (Ito et al., 1989), and evaluated its performance consideration is given to improving the directivity of
in the International Sodar Inter-comparison acoustic array antennas by choosing a better arrange-
Experiment (ISIE) at Boulder Atmospheric Observa- ment of electro-dynamic transducer elements to have
tory (BAO) in 1988 (Gaynor et al., 1990). Instead of specified aperture and directivity. The newly-
three parabolic reflectors in a tri-monostatic Doppler designed array antenna is tested by making compar-
sodar, the phased array system can make five direc-
isons with tower instruments making wind measure-
tional acoustic beams (i.e. a vertical beam and four
ments.
steering beams to the north, south, east and west)
which are synthesised from the signals with phase shifts
in an electro-dynamic speaker array. The horizontal 2. Outline of the system
mean winds derived from two pairs of beams in oppo-
site directions contain less influence of beam refraction Figure 1 shows the array antenna used in this study.
than those derived from three beams and less influence The acoustic sources consist of 216 low-cost transducer
of misalignment of the level for antenna installation. In elements, and the array aperture is contained in a
addition to these advantages, a five-beam system has square with 150 cm sides. The 16×16 array of elements
the additional information to correct measurement are triangularly cut off by ten elements in each of the
errors, and provides the wind variances with the com- four corners to make the symmetric beam. The spacing
pensation of such measurement errors. The details of length d between the elements is about 0.6λ, with λ
them are discussed by Ito (1997). being the wave length at the operating frequency of 2.1
kHz. Each element is an electro-dynamic speaker
In the ISIE, comparisons of wind speed between the which has a folded-horn diameter of 90 mm and a rated
sodar and sonic anemometers on the BAO tower power of 5 W. The sound pressure level at the fre-
showed the rms differences to be 0.89 m s–1 (Takehisa quency range of 2.1 to 2.7 kHz is better than 100 dB
et al., 1991). The 5×5 array system used in the ISIE W–1 at 1 m. Variations in the sound pressure level are
was, however, sometimes disturbed by fixed echoes. within 2 dB of each other, but some elements go out of
In another case, a commercially available phase array tolerance by up to 40–50 degrees on the phase lag,
149
14698080, 1998, 2, Downloaded from https://rmets.onlinelibrary.wiley.com/doi/10.1017/S1350482798000681 by Algeria Hinari NPL, Wiley Online Library on [01/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Y Ito
transmitting power and receiving sensitivity toward the
centre of the array is achieved by assigning a weighting
factor to each row or column of elements. If we apply
a Chebyshev polynomial to the weighting factor, the
synthesised acoustic field is optimised to suppress side-
lobes. This is the so-called aperture tapering. The 16-
channel transmitter/receiver signals have a phase shift
of 74 between adjacent channels to steer the main lobe
to 20 from the zenith. The hardware used in the system
is described in Ito et al. (1995). The preliminary exper-
iment has been carried out under the specifications
summarised in Table 1.

Table 1. Specifications of the phased array Doppler


sodar

Parameter Value
Operating frequency 2100 Hz
Transducer array 16×16 (216 elements)
Nominal output power 600 watts (electric)
Antenna aperture 2.1 m2
Zenith angle of steering beams 20°
Figure 1. Photograph of the phased array antenna (top view).
Beam width 10°
Beam arrangement 5 beams
therefore it is necessary to select the element we use. (4 orthogonal, 1 vertical)
Pulse length > 10 m s (programmable)
Figure 2 shows the radiation pattern of the transducer
Pulse repetition period > 2 s (programmable)
element compared with the approximated radiation
Doppler frequency estimation FFT
from a spherical source of 90 mm in diameter. This
shows that the spherical approximation exactly repre-
sents the radiation within 60° of the centre of the beam
and the practical sidelobe radiation is not more than 3. Design and evaluation of beam pattern
that estimated with the spherical radiation.
The principle of beam formation by an array antenna in
To make five directional beams, the 16-channel trans- a radar system has been studied for many years (e.g.
mitting signals with phase shifts are provided to the Skolnik, 1962). An application to the acoustic remote
row or column transducer elements. Increasing the sensing has been discussed in the following papers:

Figure 2. Radiation pattern of the transducer element; the measured radiation pattern (A), the polynomial approximation of
the measured pattern (B) and the approximated radiation from a spherical source of the same diameter (C).

150
14698080, 1998, 2, Downloaded from https://rmets.onlinelibrary.wiley.com/doi/10.1017/S1350482798000681 by Algeria Hinari NPL, Wiley Online Library on [01/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Acoustic antenna for a phased array Doppler sodar
Stephanou & Mavrakis (1986), Ito et al. (1989), Bradley Figure 4. Following the conventional sodar system,
& Roberts (1990), Liaskas et al. (1994) and Khanna & which has been used with the beam width of about 10°
Sharma (1996). Here, the author examines the charac- (Kaimal et al., 1980), an aperture diameter D (7Md)
teristics and the optimum arrangement under con- larger than 7λ/cos θ0 is recommended. For the antenna
straints inherent in the acoustic array: for example, the with aperture tapering mentioned in the following, we
finite aperture of the transducer element being larger need a somewhat larger D of about 9.5λ/cos θ0 to make
than a half wavelength, the directivity of the element this beam width.
and the effect of an acoustic shield. The directivity
function R(θ) of the radiation pattern is defined by the To suppress the sidelobe radiation, aperture tapering
product of the element factor Re(θ) generated from an with a Chebyshev distribution is employed in the pre-
individual source element and the array factor Ra(θ) , sent system. It gives the optimum aperture distribution
the latter being the pattern of an array composed of which minimises the beamwidth for a given sidelobe
isotropic elements. level; conversely it produces the lowest sidelobe level
when the beamwidth is specified. The details of this
2 J1 ( χe ) sin Mχ a
R (θ ) = Re (θ ) ⋅ Ra (θ ) = ⋅ (1) technique are found in the literature (e.g. van der Maas,
χe M sin χ a 1954). The aperture tapering is carried out by giving the
where gradation of transmitting power and receiving sensitiv-
χe = π De (sin θ − sin θ0 ) / λ, χ a = πd (sin θ − sin θ0 ) / λ ity to the transducer elements for each row or column
of the array.
J1 is the first-order Bessel function, De the aperture of
In an even symmetric array, the acoustic sources with
an element, θ the angle from the zenith, θ0 the steering
the weighting factors of amplitude W(m) form the
angle of the main lobe, λ the wavelength, and d the
directivity, as follows:
spacing length between the elements. Besides the direc-
tivity of each element determined by Re(θ), equation (1) M /2
1
shows that the directivity depends on the ratio d/λ and Rat (θ ) =
2W0
∑ W (m ) {exp [ j ( 2m − 1) χ a ] + ( 2)
the number of acoustic sources M. m =1

exp [ − j ( 2m − 1) χ a ]}
Figure 3 indicates the directivity function R(θ) for M =
5, 10, 16, where Re(θ) is approximated by the radiation where
from spherical sources of χe = χa. We can design the M /2
beam pattern (i.e. the number of sidelobes, these side- W0 = ∑ W (m )
m =1
lobe levels and the beam widths) in each multiple point
source from this figure. Even though a large value of M,
inevitable with the large aperture, makes intensive and Rat(θ) denotes the array factor for aperture
directivity, the large sidelobe appears at χa = π; there- tapering. If we apply a Chebyshev polynomial to
fore, it is necessary to choose χa < 2.5, i.e. d/λ <
0.8(1+sin θ0)–1. This means that we should choose the
relation d/λ < 0.8 for a vertical beam and d/λ < 0.6 for
a steering beam of θ0 = 20o . Similarly the beam width
Ψ, defined as the double angle of R1/2(θ) being reduced
to half of its centre value depends on the number of ele-
ments M and the acoustic spacing d/λ as shown in

Figure 4. Beam width Ψ as a function of the product of the


Figure 3. Directivity function R(θ) synthesised from the mul- number of source elements M and the acoustic spacing d/λ .
tiple point sources. Each point is assumed to be a spherical The arrows indicate the value of M d/ λ which provides a
source of χe= χa . beam width of 10° for no taper and tapered cases.

151
14698080, 1998, 2, Downloaded from https://rmets.onlinelibrary.wiley.com/doi/10.1017/S1350482798000681 by Algeria Hinari NPL, Wiley Online Library on [01/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Y Ito
Table 2. An example of the weighting factor of amplitude W(m) required to suppress the sidelobe by –35 dB accord-
ing to an eight-stage Chebyshev distribution

Edge Centre
m 8 7 6 5 4 3 2 1
8 7 6 5 4 3 2 1
W(m) 0.172 0.245 0.384 0.540 0.697 0.838 0.943 1

approximate the summation term in equation (2), the nearly symmetric in space. Figure 6 shows the beam
weighting factor W(m) can be chosen to reduce the patterns with and without aperture tapering computed
maxima in the range 0< χa< π in Figure 3. An example from a planar array composed of 216 elements. We can
of the weighting factors calculated for the eight-stage expect that the aperture tapering mentioned here sup-
gradation is listed in Table 2. These factors are presses the sidelobe level by 10 dB below the non-
designed to suppress the sidelobe by 35 dB below the tapered basic beam pattern. To make sure the computa-
centre of the beam. The resultant beam pattern is tion of the directivity function is accurate, the antenna
shown in Figure 5, where the element factor Re(θ) is gain pattern has been checked experimentally. The
given by the measured radiation. Even though the measured radiation beam pattern is plotted in Figure 6,
beam width becomes broad and the transmitted total where the beam is designed to steer at 20o from the
power decreases by a factor of 0.6 in the present case, zenith. The radiated acoustic pressure was measured by
Figure 5 shows that the sidelobe radiation with aper- a condenser microphone which was kept at a distance
ture tapering is more suppressed than that without of 4 m from the array to avoid interference in the region
tapering or in comparison with the radiation from a of the acoustic near field. Each plot is averaged over
spherical source of the same size in diameter. The 5×5 several data points, since there is some scatter in the
array tested in the ISIE has a broad main lobe and large measured signals within 2 dB. Although the calibrated
sidelobes. main lobe in Figure 6 is somewhat broad and the beam
tilt shows 21o rather than the expected value of 20o, the
The ideas discussed above have produced a phased resulting data support the current acoustic design. The
array Doppler sodar which has an antenna using a measured radiation shows that the sidelobe level
16×16 array with 216 elements. Cutting off ten ele- around the main lobe is less than –30 dB. This agrees
ments in each of the four corners can make the beam with the computation. Even though it was difficult to

Figure 5. Directivity function of 16 row or column array: without aperture tapering (A) and with tapering (B), in comparison
with a spherical radiation (C). Directivity function for a 5×5 array without aperture tapering is also shown (D).

152
14698080, 1998, 2, Downloaded from https://rmets.onlinelibrary.wiley.com/doi/10.1017/S1350482798000681 by Algeria Hinari NPL, Wiley Online Library on [01/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Acoustic antenna for a phased array Doppler sodar

Figure 6. Measured radiation pattern in comparison with the calculated patterns formed by the planar arrays with and with-
out aperture tapering composed of 216 source elements.

get accurate measurements, the sidelobe level in the aperture is 9.6λ in width and the size of the acoustic
horizontal direction (θ = 90o) was less than –40 dB. shield is 16.8λ in upper aperture and 6λ in height. The
Reduction of the horizontal radiation is important to farfield gain pattern is computed for various heights of
avoid fixed echoes and to prevent acoustic leakage. the shield, as listed in Table 3. The estimation repre-
Despite the difficulties of making accurate measure- sented in Figure 7 shows that the acoustic shield is use-
ments of acoustic intensity in the field, the measured ful for the array antenna. For the case of 6λ, we can
beam pattern in Figure 6 confirms that the sidelobe expect a sidelobe reduction of 10 dB compared with the
suppression has been achieved. value without the shield for a vertical beam and of 5 dB
for a steering beam of θ0 = 20° in the horizontal direc-
In practice, we use an acoustic shield around the tion. As shown in Figure 7, a shield higher than 12λ
antenna aperture to enhance the sidelobe suppression. impairs the function of the antenna, because the main
The effect of the acoustic shield is estimated by refer- lobe of the tilted beam becomes asymmetric and
ring to the numerical study for a circular antenna given decreases in level relative to the sidelobes. The tilt of an
by Strand (1971) or Adekola (1976). In the case of the acoustic shield larger than the beam tilt seems to keep
array antenna, the following expression is assumed to the synthesised beam shape.
give the directivity function R(θ).

(1 + cos θ ) 4. Comparison experiment


R (θ ) =
2
∑ pi exp [ − j 2 π xi sin θ / λ ] /
i
The phased array Doppler sodar with the above-men-
(1 + cos θ0 )
2
∑ pi exp [ − j 2 πxi sin θ0 / λ ] tioned design was tested using the tower instruments
for comparison. An experiment was held at the
i

where Meteorological Research Institute (MRI) of Japan in


216 July 1995. Sonic anemometers were mounted on the
Pi = ∑ A ( n) exp [ jxn sin θ0 ] 213 m instrumented tower at three levels: 100, 150 and
n =1 200 m. The sodar system was located about 350 m to
exp [ j 2 π (di −n − h) / λ ] γ (di −n )
Table 3. Height and tilt of an acoustic shield used in the
pi is the acoustic pressure of the ith finite area in numerical estimation. Tilt is calculated from the sizes of
the upper aperture at a height h, A(n)exp[jxnsin θ0] 9.6λ in the lower and 16.8λ in the upper aperture of the
is the acoustic emission from the nth array element, shield. The underlined value is adopted in this study
and γ(di–n) is the decay of the acoustic signal with dis-
tance di–n given by Strand (1971). The factor h 0 3λ 6λ 9λ 12λ 15λ
exp[j2π(di–n–h)/λ] describes the change of phase of the
height (m) 0 0.48 0.97 1.46 1.94 2.43
acoustic signal over distance di–n. The array antenna tilt (°) 50 31 22 17 13
used in the present study has been evaluated. The array
153
14698080, 1998, 2, Downloaded from https://rmets.onlinelibrary.wiley.com/doi/10.1017/S1350482798000681 by Algeria Hinari NPL, Wiley Online Library on [01/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Y Ito

Figure 7. Estimation of the effect of an acoustic shield for (a) a vertical beam and (b) a steering beam of θ0 = 20°. Directivity
pattern is computed for various heights of the shield. A height of 6λ is practically suitable for sidelobe suppression and is adopted
in this study.

the north of the tower. The data acquisition was carried plotted in Figure 8 exclude the data for windspeeds
out every one second for all tower sensors. On the less than 1 m s–1. The results of comparisons at 150 m
other hand, the sodar was cycled through five beams and 100 m were almost the same as those at 200 m. To
once every 15 s. Four steering beams were tilted 20o examine the reliability and accuracy of the wind mea-
from the vertical to the north, south, east and west surements, three basic criteria adopted by Kaimal et
directions for the phased array sodar. The 10 min aver- al. (1984) are used for the present analysis. These cri-
aged windspeed was compared between the sodar and teria are the bias b, the rms difference (the compara-
the sonic anemometer at each level. bility) c, and the standard deviation of the difference
(the precision) p. The results of these criteria and the
Figure 8 shows the scatter diagrams of wind speed and correlation coefficient R are summarized in Table 4.
direction at the 200 m level. The wind direction data They are compared with those measured by the 5×5
154
14698080, 1998, 2, Downloaded from https://rmets.onlinelibrary.wiley.com/doi/10.1017/S1350482798000681 by Algeria Hinari NPL, Wiley Online Library on [01/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Acoustic antenna for a phased array Doppler sodar
are generally better than those obtained in the previ-
ous experiment (Gaynor & Baker, 1991). Though the
local topography of the BAO is flatter than that of the
MRI, the degree of agreement between the sodar and
tower data obtained in the MRI is better than in the
BAO. As the data processing is the same in both sys-
tems shown in Table 4, such difference seems to be
dependent on the directive performance of the
antenna.

5. Conclusions
The beam formation method employed in the new
phased array Doppler sodar has improved the beam
directivity by making a better arrangement of the trans-
ducer elements and using the aperture tapering tech-
nique. Evaluation shows that:

(a) an aperture diameter D larger than 9.5λ/cos θ0 for


the tapered array antenna provides a main beam
that is narrower than 10o,
(b) acoustic spacing d/λ between transducer elements
of less than about 0.8λ (1+sin θ0)–1 obviates the large
sidelobe,
(c) aperture tapering based on the eight-stage
Chebyshev distribution can suppress the sidelobe
level by 10 dB below the non-tapered basic beam
pattern.

These characteristics expected from the numerical


evaluation were confirmed in the operational field test.
An acoustic shield with a suitably designed aperture
and height will reduce the sidelobes by 5–10 dB in the
horizontal direction. The intercomparison shows that
the precision of the wind data compared with the tower
instrument is as good as or even better than that of the
previous experiment.

Figure 8. Scatter plots of (a) 10 min averaged wind speeds Acknowledgement


and (b) directions from the phased array Doppler sodar versus
the sonic anemometer at the 200 m level. R denotes the corre- The author is much indebted to Prof. Y. Mitsuta of
lation coefficient and Stv is the rms difference. Kyoto University and Dr T. Hanafusa of the
Meteorological Research Institute for valuable discus-
array system in the ISIE. Statistical values obtained in sion and help with the experiments. The comparison
the MRI are 0.14–0.28 m s–1 for b, 0.54–0.57 m s–1 for experiment was conducted by the MRI in joint research
c and 0.47–0.52 m s–1 for p at three levels. These values with the Nuclear Safety Research Association.

Table 4. The statistics of intercomparison between the sodar and tower wind speed

Statistic MRI (1995) 16×16 array BAO (1988) 5×5 array


height (m) height (m)
100 150 200 100 150 200
N 394 376 352 463 427 344
b (m s–1) 0.14 0.27 0.28 –0.74 –0.59 –0.49
c (m s–1) 0.54 0.54 0.57 1.14 1.14 0.91
p (m s–1) 0.52 0.47 0.50 0.87 0.98 0.77
R 0.96 0.96 0.96

155
14698080, 1998, 2, Downloaded from https://rmets.onlinelibrary.wiley.com/doi/10.1017/S1350482798000681 by Algeria Hinari NPL, Wiley Online Library on [01/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Y Ito
References Kaimal, J. C., Gaynor, J. E. Finkelstein, P. L. Graves, M. E. &
Lockhart, T. J. (1984). An evaluation of wind measure-
Adekola, S. A. (1976). Toward a more general integral for- ments by four Doppler sodars. BAO Rep. No.5, Wave
mulation of the pressure field of an echosonde aperture Propagation Laboratory, NOAA/ERL, Boulder, CO, 110
antenna. J. Acoust. Soc. Am., 60: 230–239. pp.
Beran, D. W. & Clifford, S. F. (1972). Acoustic Doppler mea- Khanna, R. M. & Sharma, O. (1996). Design and study of an
surements of the total wind vector. Second Symp. on acoustic antenna for a phased array Doppler sodar. In
Meteorological Observations and Instrumentation, San Proceedings of Eighth International Symp. on Acoustic
Diego, CA, Am. Meteorol. Soc., 100–110. Remote Sensing (M. A. Kallistratova, ed.), Moscow, Russia,
Bradley, S. G. & Roberts, N. L. (1990). Design and evaluation 3.67–3.71.
of a compact Doppler acoustic sounder. In Proceedings of Liaskas, T. J., Stephanou, G. J. & Martakos, D. (1994). Beam
Fifth International Symp. on Acoustic Remote Sensing (S. P. pointing errors of phased arrays for digital switched beam
Singal, ed.) Tata McGraw-Hill, New Delhi, India, 141–146. acoustic radars. Int. J. Remote Sensing, 15: 333–344.
Gaynor, J. E., Baker, C. B. & Kaimal, J. C. (1990). The inter- Skolnik, M. I. (1962). Introduction to radar systems.
national sodar intercomparison experiment, acoustic McGraw-Hill Book Company, 294–329.
remote sensing. In Proceedings of Fifth International Symp. Stephanou, G. J. & Mavrakis, D. H. (1986). Switched beam
on Acoustic Remote Sensing (S. P. Singal, ed.) Tata planar arrays suitable for high resolution acoustic
McGraw-Hill, New Delhi, India, 67–74. sounders. IEEE Transactions on Geoscience and Remote
Gaynor, J. E. & Baker. C. B. (1991). Fine time-scale compar- Sensing, GE-24: 745–750.
isons between Doppler sodar and sonic anemometer- Strand, O. N. (1971). Numerical study of the gain pattern of
derived winds. In Proceedings of Seventh Symp. on a shielded acoustic antenna. J. Acoust. Soc Am., 49:
Meteorological Observations and Instrumentation, New 1698–1703.
Orleans, LA, Am. Meteorol. Soc., 401–404. Takehisa, M, Ito, Y. & Mitsuta, Y. (1991). Precision and rela-
Ito, Y. (1997). Errors in wind measurements estimated by tive accuracy of a phased array Doppler sodar. In
five-beam phased array Doppler sodar. J. Atmos. Oceanic Proceedings of Seventh Symp. on Meteorological
Technol., 14: 792–801. Observations and Instrumentation, New Orleans, LA,
Ito, Y., Kobori, Y., Horiguchi, M., Takehisa, M. & Mitsuta, Am. Meteorol. Soc., 405–408.
Y. (1989). Development of wind profiling sodar. J. Atmos. van der Maas, G. J. (1954). A simplified calculation for
Oceanic Technol., 6: 779–784. Dolph-Tchebyscheff arrays. J. Appl. Phys., 25: 121–124.
Ito, Y., Uzawa, T., Hanafusa, T. & Mitsuta, Y. (1995). Vogt, S. & Thomas, P. (1994). Test of a phased array sodar by
Development and operational characteristics of a phased Intercomparison with tower data. J. Atmos. Oceanic
array Doppler sodar. In Proceedings of Ninth Symp. on Technol., 11: 94–102.
Meteorological Observations and Instrumentation, Vogt, S., Beyrich, F., Kalthoff, N. & Weisensee, U. (1994).
Charlotte, NC, Am. Meteorol. Soc., 507–511. Comparison of two Doppler sodar systems with a
Kaimal, J. C., Baynton, H. N. & Gaynor, J. E. (1980). Low radiosonde system during the SANA 1993 field experi-
level intercomparison experiment, instruments and observ- ment. In Proceedings of Seventh International Symp. on
ing methods. Rep. No.3, WMO, Geneva, Switzerland, 191 Acoustic Remote Sensing (W. D. Neff, ed.), Boulder, CO,
pp. 3.145–150.

156

You might also like