Laboratory and field inter-comparison of sonic and cup anemometers in Hong Kong

Transcription

1 Laboratory and field inter-comparison of sonic and cup anemometers in Hong Kong CHAN Ying-wa and TAM Kwong-hung Hong Kong Observatory 134A Nathan Road, Tsim Sha Tsui, Kowloon, Hong Kong, China Tel: , Fax: , Abstract The operational use of sonic anemometers in a tropical environment like Hong Kong with episodes of heavy rain and tropical cyclone is a great challenge to meteorological services in the region. The Hong Kong Observatory (HKO) has been evaluating the performance of sonic anemometers since 2013 with a view to identifying their accuracy and robustness. This paper first summarizes the wind tunnel calibration results for three different models of cup anemometers as well as one 2D model and one 3D model of sonic anemometers. The results showed very good linearity between wind speed measured by the various anemometers and the reference wind speed. The maximum errors for all anemometers were less than 10%, meeting WMO s accuracy requirement. Field inter-comparison results showed that sonic anemometers performed reasonably well during a tropical cyclone event under rainy and windy conditions. Preliminary results also showed that the 10- minute mean wind speed recorded by the 2D, 3D sonic anemometers compared well with that measured by the cup anemometer with mean absolute differences of around 0.5 m/s and 1.0 m/s respectively. 1. Introduction The Hong Kong Observatory (HKO) has been primarily using conventional cup anemometers and wind vane for the measurements of wind speed and wind direction respectively. Considerable efforts are required for maintaining the cup anemometers as their performances are usually affected by the increase of bearing friction due to ingress of dust and corrosion (WMO, 2014). The emergence of sonic anemometers has become an attractive option for measurements of wind speed as they have no moving parts and hence should have high durability and little accuracy deterioration (WMO, 2014). This should reduce maintenance works and save operational costs. Some models of sonic anemometers also provide measurements of vertical wind speed which may be beneficial to the monitoring of updrafts and downdrafts in severe weather events. In a tropical environment like Hong Kong, high winds and squalls are frequently recorded due to rainstorms and tropical cyclones. For example, during the passage of Typhoon York in September 1999, a maximum hourly wind speed of over 150 km/h and a maximum gust of more than 230 km/h were recorded at the offshore Waglan Island in Hong Kong (HKO, 1999). 1

2 Wind tunnel calibration tests were conducted for different models of cup and sonic anemometers to evaluate their performances in wind speed measurements particularly if they can meet the WMO s ±10% accuracy requirement at high wind speed. A field inter-comparison of the anemometers was also carried out at HKO's automatic weather station (AWS) at Tai Mo Shan (TMS) since November The TMS station is located on the highest hilltop in Hong Kong at around 955 m above the mean sea level. The objective of the comparison is to evaluate if the sonic anemometers are robust enough to be deployed in the field for operational use. 2. Models of cup and sonic anemometers Wind tunnel calibration was carried out for three existing HKO s operational models of cup anemometers, viz., Met-One SS-201, Munro IM124 and Adolf Thies as well as one 2D model WindObserver II and one 3D model WindMaster Pro sonic anemometers. A summary of the technical specifications of these anemometers is shown in Table Wind tunnel calibration of anemometers 3.1 Hong Kong University of Science and Technology (HKUST) In December 2012, the three operational models of cup anemometers mentioned in Section 2 above were installed inside a wind tunnel operated by the HKUST for comparing with the reference wind speed from a calibrated pitot tube in the range of about 5 m/s to 27 m/s (Figure 1a-c). The cross-section of the wind tunnel s test section was around 3m in width and 2m in height. The blockage caused by the anemometer set-up was less than 3% of the cross-sectional area, meeting the 10% requirement of ISO (ISO , 2007). The wind speed calibration results (Table 2) showed very high linearity with correlation coefficient exceeding 0.99 (Figure 2). The maximum error was around 1.5 m/s at a maximum reference wind speed of about 27 m/s. The errors were all less than 10%, meeting WMO s accuracy requirement on wind speed measurement. 3.2 WMO Regional Instrument Centre The three operational models of cup anemometers were brought to the Japan Meteorological Agency s (JMA) Regional Instrument Centre (RIC) at Tsukuba in January 2015 for extending the calibration range up to a maximum speed of around 90 m/s. The opportunity was also taken to calibrate the two sonic anemometers mentioned in Section 2 above to assess if their wind speed measurement accuracy could meet the WMO s ±10% requirement particularly at high wind speed. The wind tunnel operated by JMA provided two different sizes of test section with effective calibration areas of around m 2 and m 2 respectively (Figure 3a). The anemometers were placed in the appropriate test section so that the blockage of the anemometer set-up was less than 10% for cup anemometers and 5% for sonic anemometers (Figure 3b-f), meeting respectively the ISO (ISO , 2007) and ISO (ISO 16622, 2002) requirements. The calibration results are shown in 2

3 Table 3 as well as Figures 4 and 5. Again, there was a very high linearity between the values of wind speed measured by the JMA s reference standards (ultrasonic current meter and differential pressure gauge in the range of 0-25 m/s and m/s respectively) and those recorded by the cup and sonic anemometers. The associated correlation coefficients all exceeded The maximum wind speed errors were less than 5 m/s for the cup anemometers. For the sonic anemometers, the maximum error was 2 m/s for WindObserver II and less than 1 m/s for WindMaster Pro. The results showed that all the cup and sonic anemometers could meet the WMO s ±10% accuracy requirement even at high wind speed. 4. Field inter-comparison The field inter-comparison comprises two phases. In Phase I, the two sonic anemometers (WindObserver II and WindMaster Pro) as well as one type of cup anemometer (Met-One SS-201) and the wind vane (Met-One 1565C1 1 ) were installed at TMS at a height of around 2m above ground for testing qualitatively the performances of the sonic anemometers. Wind measurements were taken in several wind episodes from November 2013 to September In Phase II which started in January 2016, a new model of the 2D sonic anemometer (WindObserver 75) with similar performance but a larger wind speed measurement range of 0-75 m/s was used to replace WindObserver II. The two sonic anemometers (WindObserver 75 and WindMaster Pro) were installed close to the existing cup anemometer (Munro IM124) and wind vane (Munro IM144 1 ) of the TMS AWS. The wind mast is about 10 m in height (Figure 6). A quantitative assessment could be made to compare the 10-minute mean winds and 3-seocnd gusts recorded by the anemometers. 1 The technical specifications of Met-One 1565C1 and IM144 wind vanes are shown in Table Phase I results 5.1 Active southwesterlies in August 2014 Hong Kong was affected interchangeably by the southwest monsoon and an anticyclone during August The former triggered showers and thunderstorms while the latter brought generally fine weather to the territory. The 1-minute mean vertical wind speed recorded by the 3D WindMaster Pro sonic anemometer showed the presence of upward winds during the periods when active southwesterlies dominated. These observations were consistent with the development of showers and rainfall recorded by the raingauge at TMS (Figure 7). The vertical wind speed as measured by the 3D sonic anemometer could thus provide some hints on the development of showers. 5.2 Tropical cyclone in September 2014 Typhoon Kalmaegi moved across the northern part of the South China Sea from 14 to 17 September 2014 (HKO, 2014). Strong to gale force winds affected Hong Kong on 16 September In this 4-day period, the sonic anemometers showed similar variations compared with that of the cup anemometer (Figure 8). Unfortunately, an equipment house upstream of the anemometers generated blockage effect, causing some discrepancies of the wind speed recorded by the anemometers 3

4 (WindObserver II was located closer to the house and thus recorded relatively lower wind speed). Nevertheless, this case showed that the two sonic anemometers performed reasonably well during rainy and windy conditions. 6. Phase II results Six-month data collected from January to June 2016 was used for comparison of the 10-minute mean wind speed and 3-second gusts recorded by the anemometers (Tables 4 and 5). Analysis of the vertical wind speed recorded by the 3D sonic anemometer was also carried out minute mean wind speed From Table 4, the six-month (January to June 2016) average of 10-minute mean wind speed recorded by the Munro IM124 cup anemometer, the WindObserver 75 and WindMaster Pro sonic anemometers were 8.69 m/s, 8.59 m/s and 8.25 m/s respectively. Taking into account all the monthly averages, Munro IM124 recorded consistently the highest mean wind speed while that recorded by WindMaster Pro was mostly the lowest. The 6-month averages of mean absolute wind speed differences between Munro IM124 and WindObserver 75, Munro IM124 and WindMaster Pro as well as WindObserver 75 and WindMaster Pro were 0.48 m/s, 1.02 m/s and 0.84 m/s respectively (Table 5). Overall speaking, the average 10-minute mean wind speed recorded by Munro IM124 was around 1% and 6% higher than those recorded by WindObserver 75 and WindMaster Pro respectively minute mean wind direction The six-month average of 10-minute mean wind direction recorded by Munro IM124, WindObserver 75 and WindMaster Pro were 145 degrees, 134 degrees and 128 degrees respectively (Table 4). There was a systematic difference of 10-minute mean wind direction recorded by the anemometers with Munro IM124 measuring a larger southerly component. The discrepancies suggested further finetuning of true north alignment of the anemometers. Nevertheless, the three anemometers in general showed rather good agreement in the variation of wind direction (Figure 9) minute maximum 3-second gusts The six-month average of 3-second gusts recorded by Munro IM124, WindObserver 75 and WindMaster Pro were m/s, m/s and m/s respectively (Table 4). Opposite to the 10- minute mean wind speed, WindMaster Pro recorded consistently the highest 3-second gusts while that recorded by Munro IM124 was mostly the lowest. The 6-month averages of mean absolute differences of 3-second gusts between Munro IM124 and WindObserver 75, Munro IM124 and WindMaster Pro as well as WindObserver 75 and WindMaster Pro were 0.83 m/s, 1.70 m/s and 1.55 m/s respectively (Table 5). On the whole, the average 3-second gusts recorded by WindMaster Pro was around 5% and 4% higher than those recorded by Munro IM124 and WindObserver 75 respectively. It was noteworthy that despite WindMaster Pro recording higher 3-second gusts, the corresponding 6-month average of standard deviation was 4.06 m/s which was smaller than that of Munro IM 124 (4.16 m/s) and slightly higher than that of 4

5 WindObserver 75 (3.92 m/s) (Table 4). WindMaster Pro seemed to record a relatively higher mean value of 3-second gusts but the corresponding variability was smaller when compared with that of Munro IM Vertical wind speed measured by WindMaster Pro sonic anemometer WindMaster Pro can also measure vertical wind speed which will be useful to the detection of updrafts and downdrafts, which are sometimes associated with the occurrences of thunderstorms. This section provides further information on the characteristics of vertical winds measured by WindMaster Pro based on data collected from January to June Monthly mean Both the monthly mean vertical wind speed and the corresponding standard deviations showed generally a rising trend from January to June 2016 (Figure 10). There was a step change of mean value from 0.39 m/s in March to 1.16 m/s in April which was around 3 times larger. The increase reflected the enhanced warming of the boundary layer as well as more turbulent flow and stronger vertical winds caused by thundery activities as the season progressed from winter to summer Shear instability On 14 January 2016, the vertical wind speed recorded by WindMaster Pro showed strong fluctuations with a maximum value reaching above 4 m/s during the night (Figure 11). Moreover, a rather regular oscillation pattern was observed in the early morning from around 01:30 a.m. to 03:30 a.m. Using this 2-hour time slot of wind data, a Fast Fourier Transform (FFT) was performed and found that there was a peak at frequency of about 1.7x10-3 Hz, corresponding to a wave period of around 10 minutes (Figure 12). After a quiet period from around 6:00 a.m. to 10:00 a.m., the vertical wind speed depicted large oscillation again but no regularity of the wave motion could be identified. The Observatory s wind profiler observations showed strong west to southwesterlies in the height range from 2km to 4km while moderate southeasterlies prevailed in the low level (Figure 13). With TMS situating at a height of around 955m above the mean sea level, vertical wind shear was expected which might generate wave motion as a result of shear instability. Bulk Richardson number (Ri B) has been employed in some studies to find out the criteria for instability to occur (Zilitinkevich and Baklanov, 2002). Ri B is specified as Ri B = g θ z v θ v [( U) 2 +( V) 2 ] (1) where g is the gravitational constant, θ v is the mean virtual potential temperature, U and V are the vertical difference of velocities and z is the height difference. As a general guidance, turbulent flow may occur if Ri B is lower than a critical value of around 0.25 (Stull, 1988). The upper-air data collected from the radiosonde launched at 00UTC on 14 January

6 was used to calculate the vertical variation of Ri B in the lower atmosphere (Figure 14). An unstable layer with Ri B dropping below 0.25 was observed near the height of 950m. This tied in pretty well with the height of TMS and suggested that the large fluctuation of vertical wind speed recorded by the WindMaster Pro sonic anemometer could likely indicate the presence of Kelvin-Helmholtz instability and trigger the occurrences of atmospheric waves. Interestingly, observations from the wind profiler on that day indicated two distinct pairs of positive and negative vertical velocity occurring in the morning from midnight to around 05:00 a.m. (Figure 15). However, the atmospheric wave motion seemed to have a larger time scale of 1-2 hours which was very different from the FFT results in Figure 12 (wave period of around 10 minutes). 7. Discussion The wind tunnel calibration at HKUST and JMA s RIC showed that all three HKO s operational models of cup anemometers, viz., Met-One SS-201, Munro IM124 and Adolf Thies , as well as the 2D WindObserver II and 3D WindMaster Pro sonic anemometers could meet WMO s ±10% accuracy requirement on wind speed measurement. Very high linearity was observed between the reference wind speed and the wind speed measured by the anemometers under calibration. The associated correlation coefficients were all higher than 0.99, significant at 5% level. The maximum error for the cup anemometers was less than 5 m/s for reference speed up to around 90 m/s. For the two sonic anemometers, the error was around 2 m/s for WindObserver II and less than 1 m/s for WindMaster Pro with a reference speed up to about 60 m/s. The latter showed consistently higher accuracy than the former. From Phase II of the field inter-comparison conducted at TMS, the average 10-minute mean wind speed recorded by the cup anemometer (Munro IM124) was higher than those measured by the 2D WindObserver 75 and 3D WindMaster Pro sonic anemometers by about 1% and 6% respectively. It has been known that the response of cup anemometers is faster for acceleration than for deceleration so that the average wind speed measured can be larger than the actual average wind speed (WMO, 2014). Also, the vertical velocity fluctuations can cause overspeeding of cup anemometer as a result of reduced cup interference in oblique flow (MacCready, 1966). Hence, the relatively slow response of cup anemometer in measuring decreasing wind trend and oblique wind flow causing slight overspeeding of the rotating cups could explain why Munro IM124 recorded generally higher 10-minute mean wind speed. This also revealed a limitation of cup anemometer. On the other hand, WindMaster Pro recorded on average higher 3-second gusts that those of Munro IM124 and WindObserver 75 by around 5% and 4% respectively. Conventionally, mechanical cup anemometer is usually not good enough to reflect rapid changing of wind speed and wind direction due to its larger inertia (Pattison, 2010). In addition, the data output rates of the three anemometers were all set at 1 second but the corresponding data sampling rates were different. For Munro IM124, the data sampling rate is 1Hz based on the configuration settings of HKO s in-house developed electronic wind speed data processing board. For WindObserver 75 and WindMaster Pro, the data sampling rates were 39Hz and 20Hz respectively according to the manufacturer s specifications (Gill Instruments, 2014; Gill Instruments, 2011). With relatively lower data sampling rate but higher inertia, it was conceivable that Munro IM124 6

7 recorded smaller 3-second gusts. However, it was not yet clear why WindObserver 75 recorded a lower 3-second gusts than that of WindMaster Pro. One possible reason was the difference of data sample volume between WindObserver 75 and WindMaster Pro. For WindObserver 75, it was only a horizontal plane between the north-south and east-west transducers, whereas for WindMaster Pro, it was a volume of space comprising three different planes associated with the three sets of lower and upper transducers. According to the equipment manual, the wind speed accuracy of WindMaster Pro can apply for wind incidence angle up to ±30 from the horizontal (Gill Instruments, 2011). In this aspect, WindMaster Pro seemed to have better performance taking account of its potential capability to measure in finer details the change of wind gusts. The vertical wind speed measured by WindMaster Pro was useful for the detection of updrafts, downdrafts and the rising of warming air in the lower atmosphere. It also provided very useful information on the changes of atmospheric stability. Coupled with other meteorological observations from wind profilers and radiosondes, etc., one might possibly evaluate the presence of atmospheric wave motion triggered by vertical shear instability. 8. Conclusion Wind tunnel calibration was conducted for three HKO s operational models of cup anemometers as well as one 2D and one 3D sonic anemometers. The results showed that all anemometers could meet the WMO s ±10% requirement on measurements of wind speed. The maximum error was less than 5 m/s for cup anemometers with reference wind speed reaching a maximum of about 90 m/s. The maximum error for the 2D and 3D sonic anemometers were 2 m/s and less than 1 m/s respectively with a maximum reference wind speed of about 60 m/s. Field inter-comparison of cup and sonic anemometers was conducted at the TMS hilltop site. In Phase I, cup and sonic anemometers were installed at a height of 2m above ground for qualitative assessment of their performances. Phase II started in January 2016 with two sonic anemometers mounting close to the existing cup anemometer of the TMS AWS. Comparison results from Phase I showed that the sonic anemometers performed reasonably well in rainy and windy conditions during the passage of Typhoon Kalmaegi in September Preliminary results from Phase II using data collected from January to June 2016 found that the 10-minute mean wind speed recorded by the cup anemometer (Munro IM124) was higher than those of WindObserver 75 and WindMaster Pro sonic anemometers by around 1% and 6% respectively. On the other hand, the 3-second gust recorded by WindMaster Pro was higher those of Munro IM124 and WindObserver 75 by about 5% and 4% respectively. In addition, the vertical winds recorded by WindMaster Pro provided very useful information on atmospheric stability including the possibility to study the wave motion caused by vertical wind shear as illustrated by the weather event occurred on 14 January

8 9. Acknowledgements The authors would like to thank their colleagues Mr. M.L. Ngan for preparing the necessary data for analysis. The authors would also like to thank Mr. K.C. Tsui for his invaluable advice and comments. 10. References Gill Instruments Ltd., (2011). WindMaster Pro Ultrasonic Anemometer User Manual. Doc. No PS Issue 7 June Gill Instruments Ltd., (2014). WindObserver 75 Ultrasonic Anemometer User Manual. Doc. No PS Issue 3 July Hong Kong Observatory, (1999). Tropical Cyclones in Hong Kong Observatory, (2014). Tropical Cyclones in ISO 16622:2002(E), (2002). Meteorology Sonic anemometer/thermometers acceptance test methods for mean wind measurements. International Organization for Standardization. ISO :2007, (2007). Meteorology wind measurement Part 1: wind tunnel test methods for rotating anemometer performance. International Organization for Standardization. MacCready, P.B. and H.R. Jex, (1964). Response characteristics and meteorological utilization of propeller and vane wind sensors. Journal of Applied Meteorology, 3: pp Pattison, A, (2010). Ultrasonic wind sensors or cup anemometers? That is the question. WindPower Engineering and Development ( Stull, Roland B. (1988). An introduction to boundary layer meteorology. Kluwer Academic Publishers. 666 pp. WMO, (2014). Guide to Meteorological Instruments and Methods of Observation, WMO-No. 8, Provisional 2014 ed., World Meteorological Organization, Geneva, Switzerland. Zilitinkevich, S., and Baklanov, A. (2002). Calculation of the Height of Stable Boundary Layers in Operational Models. Boundary Layer Meteorology, 105: pp

9 Table 1 Specifications of the cup and sonic anemometers that were taken for wind tunnel calibration and deployed for field intercomparison at Tai Mo Shan. Cup Anemometer Wind Speed Wind Direction Wind Vane Range Resolution Accuracy Range Resolution Accuracy Met-One SS m/s 0.1 m/s 0.5 m/s or ±2% of wind Met-One speed 1565C1 [3] ± 2 Munro IM m/s 0.5 m/s ±0.5% < 20.6 m/s; Munro ±1.1% > 20.6 m/s; IM144 [4] ± 1 Adolf Thies 0.3 m/s or ±2% of wind m/s 0.05 m/s speed Sonic Anemometer Wind Observer II 0-65 m/s 0.01 m/s 12 m/s ± 2 WindObserver 75 [2] 0-75 m/s 0.01 m/s 12 m/s ± 2 WindMaster Pro 0-65 m/s 0.01 m/s < 12 m/s ± 2 Note [1] WMO s accuracy requirement on wind speed measurement is 0.5 m/s for winds 5 m/s and 10% for winds > 5 m/s. [2] The 2D sonic anemometer WindObserver II was replaced by a new model WindObserver 75 (with similar performance but larger measurement range) in Phase II of the field inter-comparison of cup and sonic anemometers (please see Section 4). [3] The Met-One SS-201 cup anemometer and the Met-One 1565C1 wind vane were installed at Tai Mo Shan in Phase I of the field inter-comparison. [4] The existing Munro IM124 cup anemometer and the Munro IM144 wind vane installed at Tai Mo Shan were used in Phase II of the field inter-comparison. 9

10 Table 2 Calibration results of cup anemometers using the wind tunnel at the Hong Kong University of Science and Technology. Reference wind speed [1] Met-One SS-201 Munro IM124 Adolf Thies Wind speed measured by Met-One Error Reference wind speed [1] Wind speed measured by Munro Error Reference wind speed [1] Wind speed measured by Adolf Thies Error Note [1] A calibrated pitot tube (TPL ) was used to provide reference wind speed for calibrating the cup anemometers. 10

11 Table 3 Calibration results of cup and sonic anemometers using the wind tunnel at the Japan Meteorological Agency s Regional Instrument Centre at Tsukuba. Ref. wind speed [1] Cup Anemometers Sonic Anemometers Met-One SS-201 Munro IM124 Adolf Thies Wind Observer II WindMaster Pro Wind speed measured by Met-One Error Ref. wind speed [1] Wind speed measured by Munro Error Ref. wind speed [1] Wind speed measured by Adolf Thies Error Ref. wind speed [1] Wind speed measured by WindObserver II Error Ref. wind speed [1] Wind speed measured by WindMaster Pro Note Note Error Note [1] The Japan Meteorological Agency used an ultrasonic current meter (DA-470B) and differential pressure gauge (DPI145) to provide reference wind speed in the range of 0-25 m/s and m/s respectively. [2] The maximum wind speed outputted from Munro IM124 was around 70 m/s based on the configuration settings of the HKO s in-house developed electronic wind speed data processing board. [3] The maximum wind speed outputted from the two sonic anemometers was around 60 m/s based on the manufacturer s configuration settings of the wind sensors. 11

12 Table 4 Comparison of 10-minute mean wind speed, 10-minute mean wind direction and 1-minute maximum 3-second gusts recorded by the cup anemometer (Munro IM124 with wind vane IM144) and sonic anemometers (WindObserver 75 and WindMaster Pro) at Tai Mo Shan from January to June Munro IM124 cup anemometer WindObserver 75 WindMaster Pro Measurements Month in minute mean wind speed [1] 10-minute mean wind direction [1] (degree) 1-minute maximum 3-second gust [1] and IM144 wind vane Mean Standard Deviation Mean sonic anemometer Standard Deviation Mean sonic anemometer Standard Deviation Jan 9.18 [2] 4.46 [2] 9.04 [2] 4.27 [2] 8.76 [2] 3.79 [2] Feb Mar 8.16 [3] 3.05 [3] 7.93 [3] 2.87 [3] 8.06 [3] 2.60 [3] Apr May Jun Average (Jan-Jun) Jan 102 [2] 65 [2] 89 [2] 71 [2] 83 [2] 78 [2] Feb Mar 131 [3] 50 [3] 114 [3] 53 [3] 120 [3] 57 [3] Apr May Jun Average (Jan-Jun) Jan [2] 4.97 [2] [2] 4.73 [2] [2] 4.64 [2] Feb Mar 9.39 [3] 3.47 [3] 9.36 [3] 3.24 [3] [3] 4.20 [3] Apr May Jun Average (Jan-Jun) Note [1] Wind data was extracted for comparison only if the 10-minute mean wind speed recorded by all three anemometers exceeded 3 m/s, i.e. calm or light winds were excluded. [2] Data recorded between 23 January and noon time of 25 January 2016 was not included in comparison due to icing condition corrupting the data from the anemometers. [3] No data available from WindObserver 75 sonic anemometer between 11:13 a.m. on 18 March 2016 and the end of March. No comparison was made in this period. 12

13 Table 5 Mean absolute differences of 10-minute mean wind speed, 10-minute mean wind direction and 1-minute maximum 3-second gust recorded by the cup anemometer (Munro IM124 with wind vane IM144) and sonic anemometers (WindObserver 75 and WindMaster Pro) at Tai Mo Shan from January to June Month in 2016 Munro IM124/IM144 vs WindObserver 75 Munro IM124/IM144 vs WindMaster Pro WindObserver 75 vs WindMaster Pro Mean absolute difference of 10-minute mean wind speed [1] Mean absolute difference of 10-minute mean wind direction [1] (degree) Mean absolute difference of 1-minute maximum 3-second gust [1] Jan 0.45 [2] 0.96 [2] 0.74 [2] Feb Mar 0.55 [3] 0.92 [3] 0.73 [3] Apr May Jun Average (Jan-Jun) Jan 12 [2] 19 [2] 7 [2] Feb Mar 10 [3] 17 [3] 7 [3] Apr May Jun Average (Jan-Jun) Jan 0.63 [2] 1.42 [2] 1.23 [2] Feb Mar 0.61 [3] 2.00 [3] 1.87 [3] Apr May Jun Average (Jan-Jun) Note [1] Wind data was extracted for comparison only if the 10-minute mean wind speeds recorded by all three anemometers exceeded 3 m/s, i.e. calm or light winds were excluded. [2] Data recorded between 23 January and noon time of 25 January 2016 was not included in comparison due to icing condition corrupting the data from the anemometers. [3] No data available from WindObserver 75 sonic anemometer between 11:13 a.m. on 18 March 2016 and the end of March. No comparison was made in this period. 13

14 (a) (b) Figure 1 (c) The set up for wind tunnel calibration of cup anemometers conducted at the Hong Kong University of Science and Technology (a) Met-One, (b) Munro and (c) Adolf Thies. Figure 2 The results of wind tunnel calibration of cup anemometers conducted at the Hong Kong University of Science and Technology. High linearity was observed between wind speed measured by the calibrated pitot tube and that recorded by the cup anemometers. The maximum error was around 1.5 m/s with a maximum reference wind speed of about 27 m/s. 14

15 Wind speed measured by wind sensor under calibration (a) (b) (c) Figure 3 (d) (e) (f) The set up for wind tunnel calibration of cup and sonic anemometers conducted at the Japan Meteorological Agency s Regional Instrument Centre (a) two different test sections provided by the wind tunnel, (b) Met-One, (c) Munro, (d) Adolf Thies, (e) Wind Observer II and (f) WindMaster Pro. Met-One Munro Adolf Thies Reference wind speed from JMA's standards Figure 4 The results of wind tunnel calibration of cup anemometers conducted at the Japan Meteorological Agency s Regional Instrument Centre. High linearity was observed and the maximum errors were less than 5 m/s for all the cup anemometers with reference wind speed reaching about 90 m/s. 15

16 Wind speed measured by wind sensor under calibration WindMaster Pro WindObserver II Reference wind speed from JMA's standard Figure 5 The results of wind tunnel calibration of sonic anemometers conducted at the Japan Meteorological Agency s Regional Instrument Centre. High linearity was observed and the maximum error was 2 m/s for WindObserver II and less than 1 m/s for WindMaster Pro. WindObserver 75 sonic anemometer WindMaster Pro sonic anemometer Munro IM124 cup anemometer and IM144 wind vane Figure 6 Field inter-comparison of cup and sonic anemometers at Tai Mo Shan. The wind mast of around 10m in height (left) and a zoomed view of the cup and sonic anemometers (right). 16

17 Wind Speed 15-min Rainfall (mm) Horizontal Wind Speed Vertical Wind Speed / Rainfall (mm) Horizontal Wind Vertical Wind Rainfall Active Southwesterlies Date (August 2014) Active Southwesterlies Anticyclone Anticyclone Figure Horizontal and vertical 1-minute mean wind speeds recorded by the 3D sonic anemometer (WindMaster Pro) as well as 1-minute total rainfall recorded by the automatic weather station at Tai Mo Shan from noon of 7 August to 22 August min RF MetOne Wind Observer II WindMaster Pro Figure /09/ /09/2014 Comparison of 10-minute mean wind speeds recorded by Met-One, Wind Observer II and WindMaster Pro during the passage of Typhoon Kalmaegi in September /09/ /09/2014 0

18 Wind Direction (deg.) Cup Anemometer (Munro IM144) 180 Sonic Anemometer (WindMaster Pro) Sonic Anemometer (WindObserver 75) 90 0 April 2016 Figure 9 Variation of the 10-minute mean wind direction recorded by Munro IM144, Wind Observer 75 and WindMaster Pro in April Vertical Wind Jan Feb Mar Apr May 0.18 Jun Figure 10 Variation of the monthly mean vertical wind speed measured by the WindMaster Pro sonic anemometer from January to June 2016 (The vertical bars show the standard deviations and negative speed values highlighted in red). 18

19 Wind Speed Temperature (deg.) 5 Vertical Wind TMS_temperature Time 14 January 2016 Figure 11 Time series of the vertical wind speed measured by the WindMaster Pro sonic anemometer on 14 January 2016 (The orange line shows the air temperatures recorded by the automatic weather station at Tai Mo Shan on the same day). Figure 12 Time series of the vertical wind speed measured by the WindMaster Pro sonic anemometer from 01:30 a.m. to 03:30 a.m. on 14 January 2016 (upper panel). A Fast Fourier Transform was performed using the 2-hour time slot data and found a peak at frequency of about 1.7x10-3 Hz (red circle in the lower panel), corresponding to a period of around 10 minutes. 19

20 Figure 13 Time series of upper winds measured by the Observatory s wind profiler on 14 January Strong west to southwesterlies were observed in the height range from 2km to 4km while moderate southeasterlies prevailed in the low level. Figure 14 Vertical variation of bulk Richardson number (Ri B) (blue line) calculated based on upper-air data collected from the radiosonde launched at 00UTC on 14 January 2016 (The vertical orange line indicates a Ri B value of 0.25). There was a turbulent layer near height of 950 m as Ri B fell below

21 Time (HH:MM) Height (m) 0:09-0:59 1:09-1:59 2:09-2:59 3:09-3:59 4:09-4:59 5:09-5:59 6:09-6:59 7:09-7:59 8:09-8:59 9:09-9:59 10:09-11:00 11:09-12: Height (m) > < < :10-12:59 13:09-13:59 14:09-14:59 15:09-15:59 16:09-16:59 Time (HHMM) 17:09-17:59 18:09-18:59 19:09-19:59 20:09-20:59 21:09-21:59 22:09-22:59 23:09-23:49 = 0.0 or No Data Figure 15 Vertical velocity measured by the wind profiler on 14 January Positive and negative velocities are highlighted in paleblue/blue and orange/red colours respectively. Zero velocity or no data are shown in white colour. Two distinct positive and negative vertical velocity pairs were observed in the morning from midnight to around 05:00 a.m. as marked by red and purple rectangular squares respectively. 21