wind speed profiles measured over ground using doppler sodars

Journal of Wind Engineeringand Industrial Aerodynamics 83 (1999) 83}93

Wind speed pro"les measured over ground usingDoppler sodars

Yukio Tamura!, Kenichi Suda",*, Atsushi Sasaki#, YoshiharuIwatani$, Kunio Fujii%, Kazuki Hibi&, Ryukichi Ishibashi'

!Tokyo Institute of Polytechnics, 1583 Iiyama, Atsugi, Kanagawa, Japan"Engineering Research Center, Sato Kogyo Co., Ltd, 47-3 Sanda, Atsugi, Kanagawa 243-0211, Japan

#Nishimatsu Kensetsu, 2570-4 Shimotsuruma, Yamato, Kanagawa, Japan$College of Industrial Technology, Nihon University, Chiba, Japan

%Wind Engineering Institute, 3-29 Kandajinbocho, Chiyoda-ku, Tokyo, Japan&Shimizu Corporation, 3-4-17 Etchujima, Koto-ku, Tokyo, Japan

'Housing & Urban Development Corporation, 2683 Ishikawa, Hachioji, Tokyo, Japan

Abstract

In the wind-resistant design of buildings and structures, it is very important to accuratelyassess the design wind speed at a particular site, considering the variation in wind speed withterrain roughness. In this study, the authors' group attempts to "nd a reasonable method forestimating design wind speed for a given terrain roughness, through simultaneous windobservations at altitudes up to 420 m over two sites with di!erent roughnesses using Dopplersodars. Prior to the observations, preliminary measurements were conducted to determine thee!ect of measurement conditions such as the transmission method, pulse length, power andfrequency, on both the data collectable rate and the measured values. The quality of the dataobtained under various measuring conditions was studied, and then a proper measuringcondition was determined. Then, simultaneous observations of atmospheric boundary layerwind speeds were conducted by Doppler sodars at a `seasidea area and a `suburbana area. Thecharacteristics of the mean wind speed pro"les, which were evaluated for each distinguishedwind speed level in each wind direction, are presented. The longitudinal mean wind speedpro"les in the same storms at the two sites were also compared to study the variation inlongitudinal mean wind speed as this is a!ected by inland terrain roughness ( 1999 ElsevierScience Ltd. All rights reserved.

Keywords: Mean wind speed pro"le; Doppler sodar; Terrain roughness

*Corresponding author. Tel.: #81-046-241-2174; fax: #81-046-241-2176.E-mail address: [email protected] (K. Suda)

0167-6105/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved.PII: S 0 1 6 7 - 6 1 0 5 ( 9 9 ) 0 0 0 6 3 - X

1. Introduction

It is very important to accurately estimate wind speeds in wind-proof structuraldesign. To assess the design wind speed at a particular site, it is necessary to determinethe variation in wind speed with terrain roughness. However, there is insu$cient dataavailable, particularly on wind speed pro"les up to altitudes of about 500 m and onvariation in wind speed with terrain roughness. Wind observations at a single site willprovide some information on the wind speed pro"le, but they will give no informationon variations in wind speed with terrain roughness. It is necessary to simultaneouslyobserve wind speeds at several sites with di!erent terrain roughnesses under the samewind storm conditions. In this study, the authors' group conducted simultaneouswind observations at altitudes ranging from 40 to 420 m over two sites with di!erentroughnesses using Doppler sodars to "nd a reasonable method for estimating designwind speed for a given terrain roughness.

2. Preliminary measurements

2.1. Setting up an observation with Doppler sodar

Monostatic Doppler sodars (model AR-410 by Kaijo Co., Ltd.) were used for themeasurements. The lengths of the transmitted pulses had to be increased to receivesu$cient power to collect data at higher altitudes. However, the resolution of altitudedegenerates as the pulse length is increased. In addition to single pulse transmission,double pulse transmission can be used in which higher and lower altitude regionsare probed with longer pulses and shorter pulses, respectively. The optimumpulse length was examined to obtain data up to an altitude of 500 m above theground as well as to compare the single pulse and double pulse transmissionmethods. Because the decay of sound power with propagated distance increases withfrequency, it limits the measurable altitude range depending on the noise level in thesurrounding area. Therefore, the optimum power and frequency of the pulses werealso examined by setting the power at 150, 400 and 900 W and the frequency at 1600,2400 and 3200 Hz.

2.2. Results of preliminary measurements

Preliminary measurements were conducted to determine the e!ect of measurementconditions such as the transmission method, pulse length, power and frequency, onboth the data collectable rates (DCR, hereafter), i.e., the ratio of the amount of validdata with a S/N710 dB to the total amount of data, and the measured values. Theobservation site for the preliminary measurements was in the suburban area describedin Chapter 3. An interstate highway runs along the east side of the site. Theobservations were conducted on "ne or cloudy days. Measurements were conductedfor 10 min three times under each condition.

84 Y. Tamura et al. / J. Wind Eng. Ind. Aerodyn. 83 (1999) 83}93

Fig. 1. Data collectable rates for several di!erent pulse lengths for the single-pulse transmission method(Transmitted frequency "2400 Hz, Power"900 W). (a) 500 ms; (b) 200 ms; (c) 100 ms; and (d) 50 ms.

2.2.1. Single-pulse transmission methodThe single-pulse transmission method was "rst tried by changing the pulse length

while keeping the power at 900 W and the frequency at 2400 Hz. Fig. 1 shows theDCR c of beams A and B used as inclined probes. As shown in the "gure, it isnecessary to set the pulse length longer than 200 ms to obtain data at altitudes above400 m, and shorter than 100 ms to obtain data at altitudes below 200 m. This meansthat the data at all altitudes from 40 m to 500 m cannot be obtained by the single pulsetransmission method.

2.2.2. Double-pulse transmission methodFig. 2 shows the results of measurements by the double-pulse transmission

method at various frequencies and powers. Here, the pulse length was set at 50 ms forlow altitudes and 300 ms for high altitudes. The symbols ; and p

uindicate the

longitudinal mean wind speed and its standard deviation estimated by this system,respectively. When the frequency was set at 1600 Hz, the DCR was very low ataltitudes below 100 m and above 300 m, even with the power at 900 W (Fig. 2(a)).When the frequency was set at 3200 Hz, the DCR was extremely low at altitudesabove 300 m (Fig. 2(b)). When the frequency was set at 2400 Hz, the DCR was highenough over a wide range of altitudes, particularly when the power was set at 900 W(Fig. 2(c) and (d)). However, the DCR rapidly decreased above 500 m. Based on thesedata, it was concluded that the optimal frequency was 2400 Hz for the double pulsetransmission method at 50 and 300 ms. It was also found that a higher power settingimproved the DCR.

Y. Tamura et al. / J. Wind Eng. Ind. Aerodyn. 83 (1999) 83}93 85

Fig. 2. Data collectable rates and measurements for various measurement conditions for the double-pulsetransmission method (50 ms for low altitudes and 300 ms for high altitudes). (a) Frequency "1600 Hz,power"900 W; (b) frequency"3200 Hz, power"400 W; (c) frequency"2400 Hz, power"400 W; and(d) frequency"2400 Hz, power"900 W.

3. Observations at a seaside and in a suburban area

3.1. Observation sites

Two Doppler sodars were placed at the seaside in Hiratsuka city, Kanagawaprefecture (`seasidea) and in a suburban area in the southern region of Machida city,Tokyo (`suburban areaa), as shown in Fig. 3.

The beach line at `seasidea runs from east to west and the Paci"c Ocean lies southof the site. The `seasidea has an area of condensed residential houses with somemid-rise buildings in a northerly direction. The `suburban areaa has condensedresidential areas in all the directions. Especially in the northerly direction, there isa condensed city area with many mid-rise and high-rise buildings. The `seasidea isjudged to be of roughness Category I (power law index of longitudinal mean windspeed: a"0.10) and III (a"0.20) of AIJ Recommendations [1] for southerly and


Fig. 3. Observation sites.

northerly wind, respectively. The `suburban areaa is judged to be of roughnessCategory III of AIJ Recommendations for southerly wind.

These sites are approximately 23 km apart and are located on a nearly NNE-SSWline. They are in the southern part of the Kanto plain and a condensed residential areaextends between them.

The observations began in November, 1995 at `seasidea and in April, 1995 at`suburban areaa.

3.2. Measurement and data processing conditions

Based on preliminary measurements, the double-pulse transmission method witha 10 s interval, a frequency at 2400 Hz, and a power at 900 W was adopted. The longerpulse and the shorter pulse were used for measuring at altitudes above and below100 m, respectively. The longer pulse length was set at 500 ms, because the DCRdecreased above 400 m for a 200 ms pulse, and below 200 m for a 600 ms pulse. Theshorter pulse was set at 80 ms because the noise in a strong wind was expected to bemore than that during the preliminary measurements. Measurements were taken at 20altitudes from 40 to 420 m set at nearly regular distances on a logarithmic scale.In addition to Doppler sodar, a three-cup anemometer was set to pull the trigger ateach site.

All results discussed here are 10 min-mean values. Since the measurement accuracywas not su$cient when the DCR was less than 30%, it was decided to adopt datawhich showed a more than 30% DCR at all altitudes. The vertical pro"le of10 min-mean values are examined as their ensemble averages for several referencewind speed levels.

The data were divided into 16 azimuth segments based on the reference winddirection h

R, and were categorized for every 5 m/s with the reference wind speed ;

R.

Here, the reference wind direction hR

and the reference wind speed;R

were the valuesaveraged over all measured altitudes, considering the e!ect of categorizing the recordsby wind speed levels at a particular reference altitude.

This paper shows results for the SSW and NNE wind directions. For the SSW wind,the wind blew from the ocean toward `seasidea and thence toward `suburban areaa.


Fig. 5. Pro"les of longitudinal mean wind speed for the NNE wind. (a) Seaside; and (b) suburban area.

Fig. 4. Pro"les of longitudinal mean wind speed for the SSW wind. (a) Seaside; and (b) suburban area.

Conversely, for the NNE wind, the wind blew from the city area toward `suburbanareaa and thence toward `seasidea.

3.3. Observation results

3.3.1. Longitudinal mean wind speedFigs. 4 and 5 show the pro"le of longitudinal mean wind speed ;(z) for each

reference wind speed ;R. In the legend of the "gures, the numbers of the averaged


Fig. 6. Pro"les of wind direction for the SSW wind. (a) Seaside; and (b) suburban area.

10-min data are denoted in parentheses. In the "gures, the measurements by three-cupanemometers are also plotted. The "gure shows that those plots are nearly on theextension of the longitudinal mean wind speed pro"le based on the measurements byDoppler sodars.

For the SSW wind blowing from the sea, the power law indexes a at `suburbanareaa were larger than those that at `seasidea (Fig. 4). The power law indexes arenearly constant at 0.06}0.13 at `seasidea (Fig. 4 (a)), and increased from 0.14 to 0.33 at`suburban areaa (Fig. 4(b)), with reference wind speed ;

R. The increasing trend of

power law index with wind speed at `suburban areaa is comparable with the proposalby Cook [2].

For the NNE wind blowing from the land, the power law indexes a for;R(20 m/s

at `seasidea (Fig. 5(a)) were larger than those for the SSW wind, and the indexes afor ;

R(10 m/s at `suburban areaa were larger than those at `seasidea, as for the

SSW wind.

3.3.2. Wind directionFigs. 6 and 7 show the pro"les of wind direction h(z) at altitude z de"ned as the

deviation from the reference wind direction hR

for each reference wind speed ;R.

For the SSW wind, the wind direction varied in clockwise rotation with altitude,which agrees with the Ekman Sprial rotational direction, except for very lowwind speed ;

R(5 m/s (Fig. 6). The di!erences between the wind directions at

60 and 420 m were approximately 15}203 for 5 m/s6;R(15 m/s and 103

for 15 m/s6;R(20 m/s at `seasidea (Fig. 6(a)), and approximately 53 for

5 m/s6;R(10 m/s and 15}203 for 10 m/s6;

R(20 m/s at `suburban areaa

(Fig. 6(b)).For the NNE wind, the wind direction varied in clockwise rotation with the altitude

at altitudes above 200 m at `seasidea (Fig. 7(a)). However, there was little change inwind direction with altitude z at `suburban areaa (Fig. 7(b)).


Fig. 7. Pro"les of wind direction for the NNE wind. (a) Seaside; and (b) suburban area.

Fig. 8. Pro"les of vertical mean wind speed for the SSW wind. (a) Seaside; and (b) suburban area.

3.3.3. Vertical mean wind speedFigs. 8 and 9 show the pro"le of the vertical component =(z) of the mean wind

speed for each reference wind speed ;R.

For the SSW wind, the mean wind was up-current both at `seasidea and at`suburban areaa, except for 20 m/s6;

R(25 m/s at `seasidea with few data ob-

tained (Fig. 8). At `seasidea, the vertical mean wind speed=(z) decreased with altitudez (Fig. 8(a)). At `suburban areaa, there was little change in vertical mean wind speedwith altitude (Fig. 8(b)). The vertical mean wind speed at 100 m above the ground wasabout 0.2}0.8 m/s at `seasidea and about 0.2}0.3 m/s at `suburban areaa. It may be


Fig. 9. Pro"les of vertical mean wind speed for the NNE wind. (a) Seaside; and (b) suburban area.

thought that these characteristics were caused by the upward momentum #ux due tothe roughness change from a smoother terrain to a rougher one.

For the NNE wind, the vertical mean wind was downward and there was littlechange in vertical mean wind speed with altitude both at `seasidea and at `suburbanareaa (Fig. 9). The downward wind speed increased with reference wind speed ;

Rat

`suburban areaa (Fig. 9(b)).

4. Variation in wind speed pro5le from `seasidea to `suburban areaa

It is useful to compare the pro"les of longitudinal mean wind speeds at the two sitesin the same storms to assess the e!ects of terrain roughness in between.

Our interest was focused particularly on the wind which blows from `seasidea to`suburban areaa. The SSW winds at `seasidea were therefore selected for analysis.Here, it was assumed that the storm observed at `seasidea arrived at `suburban areaaafter a delayed time. For simplicity, the delayed time was tentatively estimated as thetime necessary to move from `seasidea to `suburban areaa with the longitudinal meanwind speed at 420 m at `seasidea. The data were categorized into groups based on thereference wind speed ;

Rat `seasidea.

Fig. 10 shows the pro"les of longitudinal mean wind speed at the two sites in thesame storms. It can be recognized that due to the so-called friction, the longitudinalmean wind speeds decreased at lower altitudes after the wind blew 23 km over thefetch with the inland terrain roughness, and consequently the power law indexesa increased considerably. It was evident that the tendency was more remarkable as thelongitudinal mean wind speed increased. Thus, at `suburban areaa the longitudinalmean wind speeds at 100 m above the ground were approximately 77% and 65% ofthose at `seasidea in the case of 10 m/s6;

R(15 m/s and 15 m/s6;

R(20 m/s,

respectively.


Fig. 10. Comparison between longitudinal mean wind speed pro"les at `seasidea and `suburban areaa inthe same storms for the SSW wind for each reference wind speed;

Rat `seasidea. (a) 5 m/s6;

R(10 m/s;

(b) 10 m/s6;R(15 m/s; (c) 15 m/s6;

R(20 m/s; and (d) 20 m/s6;

R(25 m/s.

5. Conclusion

Simultaneous observations of atmospheric boundary layer wind speeds using twoDoppler sodars were conducted at altitudes ranging from 40 to 420 m over two siteswith di!erent terrain roughnesses. The two sites were a seaside area and a suburbanarea, approximately 23 km apart. The longitudinal mean wind speed pro"les in thesame storms at the seaside and in the suburban areas were also compared toinvestigate the variation in longitudinal mean wind speed pro"le due to friction withthe inland terrain roughnesses. Preliminary measurements were conducted prior tothe observations to determine the optimum measurement condition.


The following facts were con"rmed.(1) The single pulse transmission method is inappropriate for obtaining data at

altitudes from 40 to 500 m. The data at these altitudes can be obtained by the doublepulse transmission method at 50 and 300 ms. When the optimal frequency was2400 Hz, a high power setting improved the data collectable rate.

(2) For the SSW wind from the sea, the power law indexes a in the suburban area,which is 23 km from the seaside area, were larger than that at the seaside area. In thesuburban area, the power law indexes increased with longitudinal mean wind speed at100 m above the ground.

(3) The longitudinal mean wind speeds decreased at lower altitude after the windblew 23 km over the fetch with the inland terrain roughness, and consequently thepower law indexes a increased considerably. Thus, in the suburban area, the longitudi-nal mean wind speed at 100 m above the ground was approximately 77 and 65% ofthose in the seaside area for wind speeds of 10}15 m/s and 15}20 m/s, respectively.

In addition to the mean wind speed pro"les, a method of estimating the root-mean-square wind speed from the data obtained by Doppler sodars has been investigated bythe authors' group. Discussion on this item shall be included in another paper.

Acknowledgements

This research has been conducted as A JOINT RESEARCH ON ATMOSPHRICBOUNDARY LAYERS since September 1994. This report was compiled on the basisof discussions at the group meeting. The members of the group were: Y. Tamura(Tokyo Institute of Polytechnics), T. Iwatani (Nihon University), K. Fujii, O.Nakamura, K. Miyashita, K. Fujinami (Wind Engineering Institute), Y. Kobayashi,A. Sasaki, R. Sasaki (Nishimatsu Construction), K. Suda (Sato Kogyo), R. Ishibashi(Housing & Urban Development Corp.) and K. Hibi (Shimizu Corp.).

References

[1] AIJ Recommendations for Loads on Buildings, Architectural Institute of Japan, 1993 (English transla-tion, 1996).

[2] N.J. Cook, The Designer's Guide to Wind Loading of Building Structures, Vol. 1, Butterworths,London, 1985.


wind speed profiles measured over ground using doppler sodars

Documents