Professional Documents
Culture Documents
by a Wind Profiler
By Yuji Ohsaki
and
Yoshihisa Masuda
Abstract
The wind profiler developedby the Communications Research Laboratory and operating at the Okinawa
Radio Observatory is a Doppler radar for measuring the vertical profile of the three-dimensional wind
vectors from the echo signal scattered by clear air turbulence. Vegetation motion caused by wind, however,
will produce signals cluttering the Doppler spectrum and making it difficult to identify the echo signal in
the spectrum. A software-based clutter cancellation method was therefore developed, and its effectiveness
has been confirmed by comparing the horizontal wind vectors estimated from the clutter-cancelled wind
profiler data with those measured from radiosonde.
(a) (b)
Fig. 1. (a) A wind spectrum measured at 0937 LT on December 4, 1992. Azimuth and zenith angles are
respectively 225 and 10 degrees and the measurement range is 1.5 km. The dot shows the mean wind
speed estimated from a first-moment calculation. (b) Like Fig. la but with the clutter cancellation
and with 5-point running averaging.
(a) (b)
N-S comp. W-E comp.
Fig. 2. (a) A scatter-diagram of wind speeds derived from radiosonde data (SRS) and from wind-profiler
data (SWp) without the clutter cancellation for the north-south component. Positive and neg-
ative here respectively mean north-to-south and south-to-north. r and S.D. are the correlation
coefficient and standard deviation of wind speed derived from wind profiler. The equation for the
least-square-fitted line is also shown. (b) Like Fig. 2a, but for the west-east component. Positive and
negative respectively mean west-to-east and east-to-west.
tential of a clutter-cancellation method can not be to avoid errors due to interpolation. Wind vec-
investigated separately, by using data with strong tor data for comparison were obtained from Decem-
clutter at lower heights, from the effect of the sur- ber 1992 to November 1993, and the followingfour
face condition. Thus, wind profiler data at greater scatter diagrams, showing relations between wind
heights only is used. Wind vectors estimated from speed derived from radiosonde data (SRS) and wind
wind-profiler data at heights of 1.5, 1.95, and 2.4km speed derived from wind profiler data (Sp),
yVcon-
can be compared directly (with no interpolation in tain about 4000 data points.
the vertical direction) with vectors estimated from Figure 2a is a scatter-diagram between SRS and
the radiosonde data. These heights were chosen 5w p for the north-south component without clut-
944 Journal of the Meteorological Society of Japan Vol. 74, No. 5
(a) (b)
N-S comp. W-E comp.
Fig. 3. (a) As for Fig. 2a, but with the clutter cancellation. (b) As for Fig. 2b, but with the clutter cancellation.
ter cancellation. Positive and negative respectively high. When the wind speed is low, the target spec-
mean north-to-south and south-to-north. Figure 2b trum overlaps with the clutter spectrum. Since a
is similar to Fig. 2a, but for the west-east compo- clutter cancellation works well in the case of low
nent, and in Fig. 2b positive and negative respec- wind speed, as shown in Figs. 3, the author's expec-
tively mean west-to-east and east-to-west. As shown tation of Section 2 will be correct.
in Fig. 2, when the absolute value of SRS becomes Data points at heights of 1.5, 1.95, and 2.4km
large, that of Sw p is sometimes less than that of are plotted in Figs. 3. When the measurement
SRS. This differenceis due to the clutter effect. Fig- height becomes appreciable, the signal-to-noiseratio
ures 3a and 3b are analogous to Figs. 2a and 2b but (SNR) becomes low. The scattering properties be-
with clutter cancellation, and they show closer cor- tween SRS and Swp at the three heights are slightly
respondence between SRS and Swp. The slope and different from each other. These differenceswill be
intercept of the equation for the least-square-fitted caused by the dependence of the SNR on height.
line in Fig. 3 are close to unity and zero, respec- Since the aim of this study is to investigate the po-
tively. Thus it is believed that a clutter cancellation tential of a clutter-cancellation method, this depen-
does not cause bias error. dence is not treated here.
Figures 3a and 3b show good correspondencebe- 4. Concluding remarks
tween SRS and SyVpin Sy1rpranging from about -2
to +2ms-1. However,outside this range, scatter- When the clutter spectrum was canceled by
ing between SRS and SWp is enhanced. These re- software-based clutter cancellation, the horizontal
sults suggest that when the absolute value of Swp wind vector estimated from the wind-profiler data
is less than 2ms-1 (low wind speed) the clutter corresponded well to that estimated from the ra-
spectrum is symmetrical, and when Swp is greater diosonde data. This clutter cancellation method is
than this (high wind speed) the asymmetry of the thus effectivein cancellingthe clutter in the estima-
clutter spectrum is enhanced. Thus, a clutter can- tion of the horizontal wind vector..
cellation workswell when the wind speed is low, and This clutter cancellation method will also be used
it does not work perfectly when the wind speed is in estimating the vertical wind speed. The verti-
high. The wind speed at ground level of the Okinawa cal wind speed is generally less than +1ms-1, so
Radio Observatory will correlate partially to that at that it will be quite close to the clutter spectrum
greater heights (about 1 or 2 km). From the above around zero wind speed. The authors believe that a
postulations, it seems that vegetation moves back clutter-cancellation method always works well when
and forth symmetrically when the wind speed is low the clutter spectrum is symmetrical.
and moves asymmetrically when the wind speed is
October 1996 T. Ohsaki and Y. Masuda 945
ウ ィ ン ドブ ロ フ ァ イ ラ に よ り観 測 さ れ た ド ッ プ ラ ー ス ペ ク ト ラ ム 中 の
ク ラ ッ タ成 分 の ソ フ トウ ェ ア に よ る 除 去
大 崎祐 次
(通信 総 合研 究所 鹿 島宇 宙通 信 セ ンター)
増 田悦 久
(通信総 合研 究所 沖縄 電 波観 測 所)