217

1 Introduction

A wind profiler is a radar that emits elec-tromagnetic waves upward to measure upperatmospheric winds. This radar has beenrecently paid attention as a useful instrumentfor upper air observations due to its highertemporal resolution and its easier automaticoperation than those of conventional balloonmeasurement.

The Communications Research Laboratory(CRL) has developed an L-band wind profilerand has been demonstrating its usefulnesssince 1991[1]. In 1993, we installed a windprofiler at the King Mongkut’s Institute ofTechnology Ladkrabang (KMITL) inBangkok, Thailand, and started radar opera-tion in collaboration with the Ministry of Postsand Telecommunications (presently the Min-istry of Public Management, Home Affairs,Posts, and Telecommunications) in a jointresearch/experiment project on the environ-

ment monitoring technology in Asia. In 1997,we also installed a wind profiler at the Nation-al MST Radar Facility (NMRF) in Gadanki, insouthern inland India, and started radar opera-tion under the auspices of the same project(Fig.1).

With rapid urbanization and industrializa-tion now underway in Asia, the issue of airpollution caused by factories and cars isbecoming more and more serious. Becausepollutants are transported and diffused bywind, it is important to observe and to monitorwinds in the atmospheric boundary layer.Wind profilers are suitable for local air envi-ronment management as they can continuous-ly observe short-term changes in wind.

In Thailand and India, the influence of theAsia monsoon (the seasonal winds) is signifi-cant, because seasonal variation in rainfall inthe dry or rain seasons may be accompaniedby variations in wind direction. The inter-annual variation in rainfall is also largely

OHNO Yuichi et al.

4-9 Long-term Observations Using L-BandWind Profiler in Asia

OHNO Yuichi, Nipha Leelaruji, Atma R. Jain, KOZU Toshiaki,

and MASUDA Yoshihisa

Wind profiler is a Doppler radar that can observe upper winds from the ground by receiv-ing atmospheric turbulence echo. The Communications Research Laboratory (CRL) devel-oped an L-band wind profiler and has been carrying out continuous wind observations atKoganei, Tokyo since 1993. Ministry of Posts and Telecommunications (presently, Ministryof Public Management, Home Affairs, Posts and Telecommunications) and CRL alsoinstalled L-band wind profilers at Bangkok, Thailand and Gadanki, India, and has been mak-ing wind observations in cooperation with the King Mongkut's Institute of Technology Lad-krabang (KMITL), Thailand and the National MST Radar Facility (NMRF), India. Seasonal anddiurnal variation of winds and classification of rain type using those wind profiler data areshown in this paper.

Keywords Wind profiler, Atmospheric boundary layer, Asia monsoon, Diurnal variation of winds,Rain type classification

218

influenced by monsoons. In 1995, the GEWEXAsia Monsoon Experiment (GAME) startedunder the Global Energy and Water CycleExperiment (GEWEX), a worldwide programof climate research. In 1998, GAME was con-ducting intensive observations in Asia. Weparticipated in this project and have providedwind data measured by the wind profilers in

Thailand and India to a number of meteoro-logical and climate researchers[2].

In this paper, we will mainly discuss theanalysis results of wind and rainfall data usingthe long-term observations obtained by the L-band wind profilers in Tokyo, Thailand, andIndia.

Journal of the Communications Research Laboratory Vol.49 No.2 2002

Observation sites of L-band wind profilersFig.1

219

2 L-band Wind Profiler System

While precipitation particles such as rainor snow are the main scattering bodies in caseof meteorological radar, the atmospheric scat-tering is used in case of wind profiler. Atmos-pheric scattering occurs due to variations inthe refractivity index caused by turbulence. Avariation in refractivity index on the order ofone half-wavelength of the electromagneticwave contributes to such scattering (calledBragg scattering)[3]. A wide range of frequen-cy bands, from the VHF to UHF, is used forobserving atmospheric scattering. Becausethe L-band wind profiler has relatively highfrequencies, it can work using a small antenna,with only a few square meters. However, itsobservation range is limited to less than a fewkilometers in altitude, as the scattering intensi-ty in the upper atmosphere is rather small.The scattering intensity varies depending onatmospheric conditions (temperature, humidi-ty, and turbulence). It is usually possible toobserve higher altitudes from sites subject tohigh temperature and high humidity, such asin Thailand and India. It is possible toobserve at an altitude ranging from eight toten kilometers in case of rain, since the scat-tering intensity from the precipitation particlesis high in the L-band.

The wind profiler observes the wind com-ponent in the direction of the radar beam bymeasuring the Doppler frequency shifts of

atmospheric echoes. Assuming horizontalhomogeneity, it is possible to measure a 3-dimensional wind vector that includes the ver-tical direction, using observations of morethan three beams.

Table 1 lists the specifications of windprofilers operated at CRL in Koganei, KMITLin Bangkok, and NMRF in Gadanki. Thetransmit frequency is 1357 MHz and the peaktransmitted power is 1 kW for all three windprofilers. The wind profilers at CRL andKMITL change their beam directions bymechanically driving the parabolic antenna,while the wind profiler at NMRF can changeits beam direction electronically using thephased array antenna. The three beams aredirected vertically, eastward, and northward,at all three observation sites. The zenith angleof the beams inclined to the east and north is 8degrees at the CRL site and 15 degrees at theKMITL and NMRF sites. Normally, the windprofilers operate 24 hours a day, and repeatobservations at time intervals of five minutesat CRL and KMITL, and at time intervals of11 minutes at NMRF, alternating between twopulse widths. Although occasional power fail-ures or mechanical problems interrupted auto-matic observation, so far we have been operat-ing the wind profilers for tens of thousands ofhours, and obtaining massive amounts of dataat each of the three observation sites.

OHNO Yuichi et al.

Main specifications of L-band wind profilersTable 1

220 Journal of the Communications Research Laboratory Vol.49 No.2 2002

Seasonal variations in horizontal wind observed by wind profiler in KoganeiFig.2

Seasonal variations in horizontal wind observed by wind profiler in BangkokFig.3

221

3 Seasonal and Diurnal Variationsin Wind Measured by Wind Pro-filers

The seasonal and diurnal variations inwind at each observation site are analyzedusing the wind profiler’s long-term observa-tion data. Figs.2, 3, and 4 show the seasonalvariations in horizontal wind at Koganei,Bangkok, and Gadanki, respectively. Theseresults are derived from wind profiler’s dataobtained over three years of observation, andaveraged over each 10-day period. The arrowsindicate the direction of wind: arrows pointingupward indicate wind blowing toward thenorth (i.e., southerly winds); arrows pointingright indicate wind heading the east (i.e., west-erly winds). The lengths of arrows are propor-tional to wind velocities. (See legends at thebottom-right corner.) The coordinates indicatethe height from the ground. Wind data aredrawn up to 3.5 km at intervals of 150 m forKoganei and Bangkok, and up to 5 km at

intervals of 300 m for Gadanki. To make iteasier to note the variations in seasonal winds,the westerly winds are indicated in red andeasterly winds in blue. In case of failure ofthe radar, or few effective observations causedby weak atmospheric echoes, no arrows aredrawn in these figures.

At Koganei, although the westerly windsare strong in winter and easterly winds occa-sionally appear from August to September, theseasonal variations are not obvious. Con-

OHNO Yuichi et al.

Seasonal variations in horizontal wind observed by wind profiler in GadankiFig.4

Average diurnal variation in horizontalwind observed by wind profiler inKoganei

Fig.5

222

versely, at Bangkok and Gadanki, distinct sea-sonal variations can be seen; westerly windsare predominant from May to September, witheasterly winds predominant from October toApril (with slight variations from year toyear). In Thailand and India, westerly windsdominate during the rain season, and easterlywinds are prevalent during the dry season.This indicates that both regions are affected bythe Asia monsoon.

Making the most of a high temporal reso-lution of wind profiler, diurnal wind variationsare analyzed. Fig.5 shows the diurnal varia-tion in horizontal wind averaged over eachhour over the past nine-year period atKoganei. Averaging the data over a long peri-od of time eliminates random factors withtime caused by the passage of fronts or low-pressure systems. Then, time-dependent com-ponents are extracted by this method. As canbe seen in the Fig.5, there are clear diurnalvariations below a height of 1 km, in whichnortherly winds dominate in the morning andsoutherly winds dominate in the afternoon.This variation comes from land and seabreezes due to the difference in heat capacitiesbetween the land and the sea. At night, the seais warm relative to the land and a land breeze(northerly breeze) thus blows from northernKanto to Sagami Bay. After sunrise, the landgradually warms up and a sea breeze (souther-ly breeze) blows from the sea to the land. Inthe afternoon, the land subsequently coolsdown and again a breeze blows from the land

toward the sea. Although it is unclear in thisfigure, at altitudes above 1 km there are com-pensation winds blowing in opposite direc-tions; southerly winds in the morning andnortherly winds in the afternoon[4].

Fig.6 shows the diurnal variation in hori-zontal wind obtained by averaging during therain season in Bangkok using the samemethod. In this figure, the diurnal variationsare not as obvious as in the figure of theKoganei data, but a slight reduction of south-westerly winds in the daytime can be seenbelow 1 km. Such diurnal variation is not dueto the land and sea breezes, but due to theupward development of the atmosphericboundary layer in the daytime, transformingthe surface drag to the upper atmosphere; thusweakening winds at altitude around 500 m[5].

While it is difficult to detect diurnal varia-tions through upper wind observation usingconventional balloons, observation using thewind profiler enables this type of analysis withits high temporal resolution. Since air pollu-tion due to automobile exhaust is particularlyserious around Bangkok, information on thediurnal variation in wind is particularly usefulin predicting high pollution area.

4 Rain-Type Classification

The L-band wind profiler can detect pre-cipitation particles such as rain or snow with ahigh sensitivity as well as atmospheric scatter-ing. Williams et al. classified rain as convec-tive rain and stratiform rain by using the verti-cal observation of the wind profiler[6]. Forstratiform rain, ice and snow fall calmly in theupper atmosphere, melting into rain beforereaching the ground. At the melting altitude,where the atmospheric temperature is around0 degrees, strong scattering echo (called the“bright band”) is observed. Stratiform rainhas low variations over time and tends to con-tinue falling for a long time. Conversely, con-vective rain develops from the collision ofprecipitation particles in a strong convection.There is no obvious bright band, its rain ratevaries over time, and the duration is relatively

Journal of the Communications Research Laboratory Vol.49 No.2 2002

Diurnal variation in horizontal wind dur-ing the rain season observed by windprofiler in Bangkok

Fig.6

223

short. This classification of rain types is notonly important for investigating the formationof clouds, but is also significant for rain ratemeasurements using radars, because it affectsdifferences in drop size distribution.

Following the methods adopted byWilliams et al., we have classified rainfall asstratiform or convective, from the wind profil-er data in Bangkok, by distinguishing the lay-ers of snow, rain, and the bright band from thescattering intensities and the vertical veloci-ties. Precipitation that have features of thebright band was classified as stratiform, andthe rest were classified as convective. At lowlatitudes, the bright band appears at an altitudebetween 4 and 5 km. Unfortunately, maxi-mum sampling altitude of the wind profiler inBangkok is occasionally below the altitude ofthe bright band, so stratiform rain is somewhatunderestimated in this statistic analysis.

Fig.7 shows the monthly statistics for eachrain type, expressed as a percentage of theduration of rainfall in the observation time.The lower and upper parts of the bar chartindicate stratiform and convective rain,respectively, where “N” indicates months withlittle or no data. As described above, Bangkokis affected by the monsoon, and the seasonalvariation in rainfall are clear: the percentageof the rainfall is 1 percent or below during the

dry season, but increases to about 10 to 15percent during the rain season. Convectiverainfall is slightly less prevalent than strati-form rainfall in the total period: the stratiformrainfall constitutes about 3 percent and theconvective rainfall constitutes about 2 percent.From 1997 to 1998, when El Nin~o occurred,the percentage of the rainfall did not exceed10 percent even during the rain season, result-ing in less rain overall. Stratiform rainfall inparticular was less in this period.

Since it is difficult to classify rainfall asconvective or stratiform using ground-basedinstruments such as rain gauges, radar obser-vations such as a wind profiler are thusrequired. We are planing to make use of thisrain type classification results for the groundvalidation of Precipitation Radar (PR) onboard the Tropical Rainfall Measuring Mis-sion (TRMM).

5 Conclusion

In this paper, we have discussed seasonaland diurnal variations in horizontal wind, andthe classification of rain type based on dataobtained through long-term observations usingthe L-band wind profilers in Koganei,Bangkok, and Gadanki. Although meteoro-logical observation data is lacking in the Asia

OHNO Yuichi et al.

Monthly proportion of each rain type in BangkokFig.7

224

monsoon regions due to financial limitations,substantial meteorological observations of allkinds are required to predict availability ofwater-resource, since this region has large sea-sonal and inter-annual variations in rainfall.To forecast monsoon behavior, information onwinds in the upper atmosphere is vital; it is

thus hoped that the wind profiler will becomewidespread in Asia, and its observation net-works are established. To this end, CRL willcontinue to contribute its accumulated dataand experience to the development of remotesensing technologies in the Asia monsoonregions.

Journal of the Communications Research Laboratory Vol.49 No.2 2002

References1 Y. Masuda, K. Okamoto, K. Nakamura, T. Ihara, T. Onishi, "Development of a lower atmospheric

wind profiler", Journal of the remote sensing society of Japan, Vol.12, pp.43-56, 1992.

2 N. Yamazaki, H. Kamahori, A. Yatagai, K. Takahashi, H. Ueda, K. Aonashi, K. Kuma, Y. Takeuchi, H.

Tada, Y. Hukutomi, H. Igarashi, H. Hujinami, Y.Kajikawa, "Release of GAME Reanalysis Data",

TENKI, Vol.47, pp.659-663, 2000.

3 S. Kato, S. Fukao, K. Hukuyama, K. Wakasugi, T. Sato, "Observation of middle atmosphere using

large-sized radar", KISHO KENKYU NOTE, Vol.144, pp.1-55, 1982.

4 Y. Ohno, "Wind changes observed by CRL wind profiler at Koganei", KISHO KENKYU NOTE,

Vol.193, pp.227-234, 1999.

5 J. Kondo, "Meteorology of atmospheric boundary layer", TENKI, Vol.46, pp.585-594, 1999.

6 Williams, C. R., W. L. Ecklund, and K. S. Gage, "Classification of precipitating clouds in the tropics

using 915-MHz wind profilers", J. Atmos. Oceanic Technol., Vol.12, pp.996-1012, 1995.

MASUDA Yoshihisa, Dr. Eng.

Former Staff of CommunicationsResearch Laboratory

Nipha Leelaruji

Assistant Professor, Department ofTelecommunication Engineering, KingMongkut’s Institute of Technology Lad-krabang

Atma R. Jain, Ph. D.

Director, National MST Radar Facility

KOZU Toshiaki, Dr. Eng.

Professor, Interdisciplinary Faculty ofScience and Engineering, ShimaneUniversity

OHNO Yuichi

Senior Researcher, Cloud ProfilingRadar Group, Applied Research andStandards Division

Meteorology, Radar Remote Sensing