SPATIAL CORRELATION AND TEMPORAL FLUCTUATIONS OF WIND VELOCITIES OBSERVED ON A PLAIN FIELD

Transcription

1 The eventh Asia-Pacific Conference on ind ngineering, November 8-12, 29, Taipei, Taiwan PATIAL CORRLATION AND TMPORAL FLUCTUATION OF IND VLOCITI OBRVD ON A PLAIN FILD Toshiaki Imai 1, Kenichi Kusunoki 2, Yoshihiro Hono 3, Tetsuya Takemi 4, Keiji Araki 1, Takaaki Fukuhara 5, Kotaro Bessho 2, hunsuke Hoshino 2 and Toru hibata 1 1 Researcher, Railway Technical Research Institute, Kokubunji Tokyo, Japan, imai@rtri.or.jp, araki@rtri.or.jp, shibata@rtri.or.jp 2 Meteorological Research Institute, Tsukuba, Japan, kkusunok@mri-jma.go.jp, kbessho@mri-jma.go.jp, shoshino@mri-jma.go.jp 3 ]ast Japan Railway Company, aitama, Japan, y-hono@jreast.co.jp 4 Disaster Prevention Research Institute, Kyoto University, Kyoto, Japan, takemi@storm.dpri.kyoto-u.ac.jp 5 ]Hokkaido Japan Railway Company, apporo, Japan, fukuhara@rtri.or.jp ABTRACT To study the time series of strong gusts and their fine-scale structure near the ground, surface meteorological observations were carried out on a plain field for 18 months. The spatial correlation of the wind velocities and the temporal wind fluctuations were investigated on the basis of the data observed simultaneously at 26 points. The coefficient of correlation between wind velocities at two points including a coastal point was found out to be larger than when an inland point was included. ith regard to temporal wind fluctuations, in spite of the flatness of the study area, the frequency of large increments in wind velocities within 5 minutes was almost the same as that of strong downslope winds observed in the mountainous area. KYORD: PATIAL CORRLATION, IND VLOCITY, TMPORAL FRUCTUATION Introduction The Japan Transport afety Board reported that Japan-Railways suffered from two railway derailment accidents caused by sudden gusts in the last four years. To maintain the safety of railways against strong winds, it is important to provide information about times and areas of these gusts. Few gusts are detected by the anemometers installed along the railway lines because gusts, including tornados and downbursts, occur locally. Therefore, the highresolution observations of gust phenomena were carried out over the honai area (Yamagata Prefecture, Japan) in order to develop a gust detection system for safe train operation. The following specialized observing platforms were used during the field campaign: the X-band Doppler radar, a network of surface weather observations, and global positioning system (GP) sondes. This paper reports the characteristics of the fluctuations of wind velocities observed by a network of surface weather observations from October 27 to March 29.

2 The eventh Asia-Pacific Conference on ind ngineering, November 8-12, 29, Taipei, Taiwan Observation sites in the honai region The network of surface weather observations was installed on the honai Plain located along the Japan ea. This area is characterized by the winter monsoon associated with enhanced cumulus cloud activities during cold air outbreaks. Twenty-six weather stations were arranged to cover the honai Plain, at intervals of 4km (Fig.1). e have measured the wind direction and wind velocity every second since October 27 at each station using the Vaisala eather Transmitter XT51. This instrument measures precipitation, pressure, temperature, and humidity at intervals of 1 s. The transmitters were mounted on the top of steel poles at a height of 5m from the ground surface. They are hereafter referred to as automated weather stations. Two X-band Doppler radars N 1km Figure 1: Distribution of automated weather stations on the honai Plain were installed near the C4 and D1 stations in order to characterize the fine-scale structure of the gusts near the ground surface, time evolution of gusts, and the associated storm dynamics along the coastal region of the Japan ea. The honai Plain is a rice field zone and is covered with the snow in winter. The roughness length of the ground surface on the honai Plain is approximately 1-1 m or less. Observational data used for analysis e analyzed the wind fluctuations by using the observational wind velocity data obtained at 26 automated weather stations from October 27 to March 29. The observational data obtained at 1 second intervals contained some noise that needs to be removed. The noise was caused by snow accretion to the ultrasonic transducers by snowstorm, breakdown of the transmitters by lightning, and writing error of the observational data to data loggers, etc. Then, after removing the noise from the data, we calculated the following statistics: 1-minute mean wind velocities, 1-minute mean wind directions, and 1-minute maximum wind velocities and wind directions when 1-minute maximum wind velocities were observed. Frequency of strong winds during observation period First, the strong winds observed on the honai Plain were statistically investigated. Figure 2 shows the relative frequency of 1-minute maximum wind velocities of 15m/s or more and 2m/s or more at each station over 18 months. Here, relative frequency means the number of 1-minute maximum wind velocities exceeding the threshold divided by the number of valid 1-minute maximum wind velocities. Coastal stations (A1, B1 and C1) experienced strong winds of and 2m/s or more. The monthly totals of relative frequencies

3 The eventh Asia-Pacific Conference on ind ngineering, November 8-12, 29, Taipei, Taiwan of the 1-minute maximum wind velocities of the 26 stations are shown in Figure 3. trong winds blew during the winter monsoon season (October to March), and a few strong winds blew during the summer season (April to eptember). Figure 4 shows the relative frequencies of the wind directions at the 26 stations when the 1-minute maximum wind velocities exceeded the thresholds. On the honai Plain, more than 8% of the strong winds of and 2m/s or more blew from the west, the west-northwest, and the northwest. Figure 5 shows examples of the relative frequency of the wind directions when the 1- minute wind velocities of and 2m/s or more were observed by the stations in Group C (C1 to C5). The reason we chose the stations in Group C was that the five stations in this Group have the following topographical features: C1 is located along the coast, C2 has the foothills on its west side, C3 and C4 are on flat ground, and C5 has a mountain on its east side. ach station showed high wind frequencies from the west, the west-northwest, and the northwest. However, the frequencies of the westerly strong winds at the C2 station were lower than those of other stations in Group C (23% of the strong winds of blew from the east-southeast relative frequency m/s or more B1 C1 A1 B2 C3 1 F1 2 A3 B3 C4 D4 D2 3 4 D3 F2 5 C5 B4 F4 C2 F3 A2 D5 D1 automated weather stations Figure 2: Relative frequency of 1-minute maximum wind velocities observed at 26 stations over 18 months relative frequency Oct-7 Nov-7 Dec-7 Jan-8 Feb-8 Mar-8 Apr-8 May-8 Jun-8 Jul-8 Aug-8 ep-8 Oct-8 Nov-8 Dec-8 Jan-9 Feb-9 Mar-9 observation period 2m/s or more Figure 3: Monthly totals of the relative frequencies of the 1-minute maximum wind velocities of the 26 stations N N NN.5 N NN N.3.1 2m/s or more Figure 4: Relative frequency of the total of wind directions at 26 stations when 1-minute maximum wind velocities exceeded the thresholds were observed at 26 stations

4 The eventh Asia-Pacific Conference on ind ngineering, November 8-12, 29, Taipei, Taiwan at the C2 station). The low frequencies of the strong westerly winds were confirmed at the D1 and 1 stations, as shown in Figure 6. This tendency was possibly attributed to the common topographical features of the C2, D1, and 1 stations. ach of three stations had the foothills on their west side. C1 N N NN.5 N NN N.3.1 C2 N N NN.5 N NN N.3.1 C3 2m/s or more C4 2m/s or more N N NN.5 N NN N.3.1 N N NN.6 N NN.5 N.3.1 C5 2m/s or more N 2m/s or more N N NN.5 N NN N.3.1 2m/s or more 1km Figure 5: Relative frequency of wind directions when the 1-minute wind velocities exceeding the thresholds were observed by stations in Group C

5 The eventh Asia-Pacific Conference on ind ngineering, November 8-12, 29, Taipei, Taiwan D1 N N NN.5 N NN N N N NN.5 N NN N.3.1 2m/s or more 2m/s or more Figure 6: Relative frequency of wind directions when the 1-minute wind velocities exceeding the thresholds were observed by the D1 and 1 stations Correlation between wind velocities at two stations Next, we focus on the spatial correlation of wind velocities observed in the honai area. Figure 7 shows the coefficient of correlation between the 1-minute mean wind velocities of any couple of the 26 weather stations versus the distance between each of the couple. It is clear that the shorter the distance between the two stations, the higher the correlation of the two stations 1-minute mean wind velocities. This tendency is clearer when the D5 and 5 stations are excluded from the 26 weather stations, as shown in Figure 7. The special characteristics of the D5 and 5 stations are that these stations are located near a mountain facing the honai area. The coefficients of correlation of wind velocities between B1 and the other 25 stations were larger than those between D5 and the other 25 stations, given the same distance between two stations (Fig.8). It is suggested that this difference was caused by variations in ground roughness, the scale of turbulence, and topography. coefficient of correlation excluding D5, 5 D5 - the other stations 1-minute mean wind velocity 5 - the other stations distance between two weather stations L (km) Figure 7: Correlations between 1-minute mean wind velocities observed at any couple of the 26 stations

6 The eventh Asia-Pacific Conference on ind ngineering, November 8-12, 29, Taipei, Taiwan coefficient of correlation B1 - the other stations D5 - the other stations 1-minute mean wind velocity distance between two weather stations L (km) Figure 8: Comparison of correlations between 1-minute mean wind velocities between two stations, including the B1 and D5 stations Characteristics of the temporal fluctuations in the observed wind velocities Finally, we investigated the increments in wind velocities at a few minutes to analyze the temporal fluctuations in wind velocities. Here we assumed the following train operation control rules: The train operation was suspended when the instantaneous wind velocity reached V m/s, and it was resumed when the wind velocity was continuously less than V m/s for over n minutes. Therefore, the approaching train can run into a section (taking m minutes to pass) when the maximum instantaneous wind velocity of the last n minutes is less than V m/s. Further, we assumed that train dispatchers watched the strong winds of instantaneous wind velocity exceeding 2m/s. Using the observational wind velocities at the 26 automated weather stations, we calculated the difference between the maximum wind velocity in the past n minutes from the base time t (hereafter u p ) and that in the subsequent m minutes after the base time (hereafter u f ), and obtained distribution of the wind velocity difference (Fig.9). The increment in wind velocity δ was denoted as the wind velocity difference, excluding negative values. Figure 1 shows the frequency distribution curve of the increments in wind velocity u δ = u f - u p u f wind velocities observed at 26 stations in when u p is greater than 15 m/s. Here, u p is the maximum wind velocity for the past 15 minutes from the base time and u f is that for the subsequent 5 minutes after the base time. The threshold of u p was set to 15m/s taking u p δ into account the frequency when u f time t becomes 2m/s or more. In Figure 1, t t the relative frequencies of the -n minutes t +m minutes increments in wind velocities δ are Figure 9: Conceptual diagram of δ larger in the winter monsoon season

7 The eventh Asia-Pacific Conference on ind ngineering, November 8-12, 29, Taipei, Taiwan than that in the summer season. To investigate whether the frequencies of increments in wind velocities are higher in the honai area than in other areas, we compared the distribution of the increments in wind velocities in the honai area with those in other areas. Figure 11 shows the increments in wind velocities observed in the honai area with those in other areas. In spite of the flatness of the study area, the frequencies of large increments in wind velocities within 5 minutes were almost the same as those of strong downslope winds observed in mountainous areas. It is indicated that the frequencies of large increments in wind velocities in the honai Plain are higher than those in other windy areas in Japan, especially in winter relative frequency summer season (from April,28 to eptember,28) winter season 1 (from Octover,27 to March,28) winter season 2 (from Octover,28 to March,29) increment of wind velocities δ (m/s) Figure 1: Relative frequency of seasonal wind velocity increment observed at 26 stations relative frequency shonai P1 offshore bridge P P3 R R2 foot of mountain increment of wind velocities δ (m/s) Figure 11: Comparison of the honai area to other areas with respect to increments of wind velocities

8 The eventh Asia-Pacific Conference on ind ngineering, November 8-12, 29, Taipei, Taiwan Conclusions The following findings were obtained by analyzing the wind data observed in the honai Plain over 18 months. 1. The correlation coefficient of the 1-minute mean velocities of any couple of 26 weather stations tends to decline with the distance between the 2 stations. 2. The coefficient of correlation between wind velocities at two stations, including the coastal B1, is larger than that when including the inland D5. 3. Large increments in wind velocities are frequently observed in the honai Plains in winter (October to March). 4. The frequencies of large increments of wind velocities within 5 minutes were almost the same as those of strong downslope winds observed in the mountainous area. Acknowledgments This research is supported by the Program for Promoting Fundamental Transport Technology Research form the Japan Railway Construction, Transport and Technology Agency (JRTT). References Takaaki F. and Toshiaki I., (27), valuation Method of Train Operation Control under trong winds Considering Temporal Fluctuations of ind Velocity, Quarterly Report of Railway Technical Research Institute, Vol.48, No.3, Kato,., H. uzuki, M. himamura, K. Kusunoki, and T. Hayashi, 27: The design and initial testing of an X- band Doppler radar for monitoring hazardous winds for railroad system. Preprints, 33rd Conf. on Radar Meteorology, Cairns, Australia, Amer. Meteor. os., CD-ROM,P13A.15. Kobayashi, F., Y. ugimoto, T. uzuki, T. Maesaka, and Q. Moteki, 27: Doppler radar observation of a tornado generated over the Japan ea coast during a cold air outbreak. J. Meteor. oc. Japan, 85(3), 321{334. Kusunoki, K., et al., 28: An overview of the honai area railroad weather project and early outcomes. Preprints, 5th uropean Conference on Radar in Meteorology and Hydrology, Helsinki, Finnland, ur. Meteor. oc., CDROM, P12.1. Toshiaki I., Taisuke., Toshishige F., 23, stimation of Actual ind peeds with Consideration of the Characteristics of Turbulence Observed in Natural inds(in Japanese), RTRT Report Vol.17,No