Reprint 1172

Use of Wind Measurements from Offshore Platforms

for TC Monitoring

Y.C. Chan & S.T. Chan

29th Guangdong-Hong Kong-Macao Seminar on

Meteorological Science and Technology,

Macao, 20-22 January 2015

Use of Wind Measurements from Offshore Platforms

for TC Monitoring

CHAN Yan-chun and CHAN Sai-tick

Hong Kong Observatory

Abstract

Wind reports from offshore platforms are crucial in filling up the

meteorological data void over the oceans. However, such reports are usually

made with anemometers located well above the sea surface and the winds

reported are significantly higher than the winds at surface due to the frictional

effect in the boundary layer. For the proper interpretation of the wind reports

in weather analysis, in particular for estimating the intensity of tropical cyclones

over the ocean, it is necessary to correct the measured wind speeds at elevated

locations to the standard 10 m level.

This paper studies the use of wind observations from offshore platforms,

including oil rigs, ships, weather buoys as well as surveillance aircraft, for

tropical cyclone monitoring. The practices and methods employed by various

centres around the world for correcting wind observations taken at different

altitudes are reviewed. Amongst these methods, the log wind profile is a

common model used to describe the wind profile within the planetary boundary

layer. The study reveals that the wind measurements from oil rigs, ships and

weather buoys, after applying log wind profile adjustment using a surface

roughness length of 0.0002 m, are highly correlated with biases less than 5%.

For a wind observation taken at a height of 100 m above sea level, the reduction

in the wind speed after correction to 10 m will be as much as 17%.

Based on limited tropical cyclone cases, the wind measurements taken by

the fixed-wing aircraft of the Government Flying Service of the Hong Kong

Government are shown to be consistent with the wind measurements collected

by other offshore platforms.

1

1. Introduction

Hong Kong, located on the southern coast of China facing the South China

Sea, is frequently visited by tropical cyclones (TCs) during the months of May

to November. Local weather can deteriorate rapidly during the approach of

TCs and close monitoring of these weather systems is essential for the issuance

of appropriate weather warnings by the Hong Kong Observatory (HKO) in a

timely manner.

The density of weather observations over the South China Sea is much

less than that on land, as it is much more difficult to set up and maintain fixed

weather stations out on the open sea. It is important to make full use of all

available observations from such stations for TC monitoring. The major

sources of observations from the ocean include ships, weather buoys,

large-scale structures on the sea like oil rigs, and derived observations from

remote sensing technology such as satellites. Although the remote sensing

technology has improved a lot during the past decades, measurements from

offshore platforms are still indispensable as a source of reliable ground truth of

the weather conditions near the surface. Since 2009, a fixed-wing aircraft

from the Government Flying Service (GFS) of the Hong Kong Government

has been commissioned to conduct surveillance flights when there is a TC

coming close to Hong Kong (Chan et al., 2011). The flight missions help to

obtain more valuable information for weather analysis and forecast.

Due to no-slip condition, wind speed near the ground or sea surface is less

than that at higher altitudes within the planetary boundary layer. For the

proper interpretation of wind reports in weather analysis, a standard height of

10 m is specified for the exposure of wind measuring instruments, with an

averaging time of 10 minutes (WMO, 2012). The instruments can, however,

be set up at different altitudes as constrained by the actual location, exposure,

or constructions around the station. By reviewing the practices at various

centres around the world for correcting wind observations taken at different

altitudes, this paper studies the use of wind observations from offshore

platforms for TC monitoring.

2. Data

The sources of wind observation data used in this study include weather

2

buoys, oil rig platforms, SYNOP SHIP reports, and the fixed-wing aircraft of

GFS. Three weather buoys operated by the Guangdong Meteorological

Bureau (GMB) over the offshore waters of Shantou, Shanwei and Maoming,

all with an anemometer installed at 10 m above sea level, are included in this

study as providing the ground truth of wind observations on the sea surface.

Observations from Panyu oil rig and Lufeng oil rig are also made available by

GMB, but the anemometers are installed at an elevated altitude at 107 m and

82 m respectively. The locations of these fixed stations of buoys and oil rigs

are as shown in Fig. 1. Wind observations are collected by the oil rigs and

weather buoys at 5- and 10-minute intervals respectively. On the other hand,

the SYNOP ship reports collected from the Global Telecommunication System

(GTS) of the World Meteorological Organization come at 3-hourly intervals.

The standard 10-minute mean wind data from the fixed stations and ship

reports are used in the analysis.

Ad-hoc surveillance flights conducted by a GFS aircraft over the northern

part of the South China Sea have been arranged during the approach of TCs to

Hong Kong. Winds were recorded on board the aircraft at 20 Hz, and

one-second mean winds were extracted for the analysis in this study. For

each surveillance mission, two flight paths were usually conducted to take

measurements, with the first leg flown at 2500 m above sea level to penetrate

into the TC, followed by the second leg at a lower altitude of about 600 m.

As the flight paths seldom pass right above any of the fixed stations, in order

to compare the winds measured by GFS aircraft with those from fixed stations,

binning of wind observations by their relative position to TC centre is needed.

In this connection, reference is made to the HKO TC best track dataset, except

for the recent tropical cyclones (that occurred in 2014) for which the

operational warning positions will be used. The TC cases covered in this

study include Severe Typhoon Vicente and Typhoon Kai-tak in 2012; Tropical

Storm Bebinca, Severe Tropical Storm Jebi and Super Typhoon Utor in 2013;

Super Typhoon Rammasun and Typhoon Kalmeagi in 2014.

The 25-km Advanced Scatterometer (ASCAT) ocean surface winds, i.e.

the derived 10-m mean winds at surface from the MetOp satellite operated by

the European Organization for the Exploitation of Meteorological Satellites

(EUMETSAT) (Verhoef and Stoffelen, 2009), are also used for comparison in

this report. ASCAT provides comprehensive spatial coverage but low

temporal frequency, with wind observations taken at most twice a day at any

3

single location during the ascending and descending passes of the satellite.

Depending on the actual scanning path of satellite, the time between

successive passes at a specific location is variable.

3. Height Correction for Winds

For the weather buoys, oil rig stations and ships, the wind measuring

instruments are located at altitude ranging from a few metres to around 100

metres above the sea surface. Several models can be used to describe the

vertical wind profiles within the planetary boundary layer. For example, the

power law relation (Kreith et al., 2010) is a common method being used for

engineering applications. For the meteorological community, the log wind

profile as recommended by Harper et al. (2010) and WMO (2012) is widely

adopted for correcting winds taken at an elevated level to the equivalent 10-m

winds.

3.1 Log Wind Profile

The log wind profile can be expressed in the form of Eq. (1), where uz, and

u* are the wind speed at height z above surface and the friction velocity

respectively. κ is the Von Karman constant (~0.4). is the stability term

that describes the atmospheric stability, d is the zero plane displacement which

accounts for the surface which is not flat but with obstacles. When there are

buildings on the ground, the mean zero vertical displacement would be

between the ground and the top of the buildings. z0 is the surface roughness

length that determines the strength of dragging of winds due to the roughness

of the surface layer. The roughness length parameter is dependent on the

terrain properties, with value for land much larger than that for open sea. L is

the Monin-Obukhov stability parameter which is positive for stable air, and

negative for unstable air.

Lzz

z

dzuu z ,,ln 0

0

*

(1)

According to WMO (2012), for offshore platforms, the reduction to

standard height can be important, but stability corrections are relatively small

which justify the logarithmic form of the reduction. In usual practice for TCs,

the stability term can be omitted by assuming neutral stability (i.e. .

The wind speed ratio between two heights (z1, z2) at the same location can then

4

be described by the division of two logarithmic functions as in Eq. (2), such

that the equivalent wind speed at 10 m can be obtained by multiplying a single

correction factor.

02

01

2

1

/ln

/ln

)(

)(

zz

zz

zu

zu (2)

3.2 Practices of Wind Correction in Use

WMO (2012) suggests that a roughness length of 0.0002 m can be used

for open sea in the log wind profile formulation (Table 1). Specifically

dealing with wind profile in TCs, Harper et al. (2010) also suggests that the TC

mean wind profile at height below 100 m is close to logarithmic and the

roughness length recommended for use for open sea conditions is between

0.0002 m and 0.005 m (Table 2).

Since the wind measurements at the three weather buoys used in this study

are taken at 10 m above sea level, no correction for height should be needed.

On the other hand, the anemometers on oil rig platforms are located at an

elevated level of about 100 m, the winds measured would be significantly

higher than those at sea surface. The practice in use at HKO is to reduce the

wind speed to 10 m by assuming a roughness length parameter z0=0.0002 m

for open sea. For the SYNOP ship reports, winds can be either measured by

anemometers, or estimated manually by observers on board the ships by

viewing the sea states. While the wind observations are corrected for ship

movement before reporting, no correction for height is being applied to the

SYNOP reports.

Practices by other centres worldwide are compared. An electronic

logbook software named TurboWin developed by the Royal Netherlands

Meteorological Institute (KNMI) adopts the log wind profile to correct wind

speed observations made on board the ships and other offshore platforms to 10

m. According to a technical report by the Met Office (Ingleby, 2009),

TurboWin is the most widely used logbook software by over 700 European,

Canadian and Australian ships in 2007. The roughness length being used in

TurboWin is 0.0016 m (Thomas et al., 2005). Meteo-France is also using the

log wind profile formula to correct wind speed observations to equivalent

10-m winds (Caroff P., personal communication 2014). Observed wind

5

speeds are corrected according to different kinds of terrain as summarised in

Table 3. The roughness length for observations on open water ranges

between 0.002 m and 0.006 m.

In general, the larger the roughness length, the stronger is the surface drag

and the more the reduction is needed for correction to 10 m from high altitudes.

The correction factor for different heights based on different roughness length

values mentioned above are summarised in Table 4 and illustrated in Fig. 2.

For a wind observation taken at height of 100 m, the corresponding correction

factor for a roughness length of 0.0002 m is 0.825, i.e. a reduction by as much

as 17%.

4. Inter-comparison among Datasets

The wind observations from oil rig platforms, weather buoys, ships, and

ASCAT are inter-compared. While the observations from weather buoys and

ASCAT are assumed to be representative of the winds at 10 m, the data

between the two datasets are first compared for verification of consistencies.

ASCAT wind data within 0.1 degree latitude/longitude of the buoys (i.e. the

nearest ASCAT observation) are extracted and compared with the buoy

observations. Data from January 2012 to September 2014 are selected for

comparison and plotted in Fig. 3. A very good correlation with R2 of 0.95 is

obtained. By performing a linear fitting of y = mx, a slope close to unity can

be obtained.

Wind observations from Lufeng and Panyu oil rig platforms are available

earlier from January 2011 onwards and data up to September 2014 are

extracted for the study. The raw wind speed observations from the two oil rig

platforms are compared with the ASCAT observations given the good spatial

coverage of the latter. The scatter plots are as shown in Fig. 4(a) and 4(b).

Again a very good correlation with R2 reaching 0.87 and 0.89 are obtained for

Lufeng and Panyu respectively. This suggests that the wind observations

taken a height close to 100 m are highly correlated with the corresponding

surface wind conditions as extracted from the ASCAT data. Wind speeds at

different altitudes are assumed to be linearly dependent, and the same linear

function of y = mx can be fitted to the data. According to Table 4, taking a

roughness length of 00002 m, a correction factor of about 0.82 and 0.84 is

needed to correct the winds taken at 107 m (Panyu) and 82 m (Lufeng) above

6

sea surface respectively. After applying the correction, the slope of the fitted

lines become much closer to one which suggests the effectiveness of the

correction [Fig 4(c) & 4(d)].

To test further the effectiveness of the correction made to the wind

observations collected by the oil rigs platforms, a comparison between the

corrected oil rig wind speed data are conducted. As the two platform are at a

distance (~200 km) away from each other, remarkably different weather

regimes could happen at the same time. In order to minimise such situations,

pairing of data are made only when the wind direction from both oil rigs are

within 5 degrees of each other. Besides, special weather situations such as

the presence of a TC locating close to the stations that would result in

significantly different wind flows at the oil rigs are excluded from the

comparison. The resulted scatter plot of the oil rig observations is as shown

in Fig. 5. A still rather good correlation with R2 of 0.61 is obtained despite

the distance between the two platforms, and the wind speed ratio is very close

to 1, suggesting that the respective correction for height applied to both oil rig

platforms are self-consistent.

Next, the wind observations from SYNOP ship reports are compared with

the fixed stations of weather buoys and oil rig platforms. Ship reports with

wind measurements taken within 1 degree latitude/longitude of the fixed

stations from January 2012 - September 2014 are extracted. Since the

observations are not exactly co-located with the fixed stations, a wind direction

difference of less than 90 degrees is taken as one criterion for pairing of data

with reasonably correlated winds. In addition to that, only wind speed data

greater than 5 m/s are chosen for the comparison to filter out those rather light

wind conditions. Similar to the previous argument, data collected during TC

situations are excluded.

While the method of observations (human/anemometer) or altitude of the

measuring instruments, if applicable, is not available in the dataset, an average

height was assumed for the wind measurements taken by ships. Based on

historical data, Thomas (2004) considered a height of 30 m to 40 m suitable

for correction of wind observations from ships. Assuming an average height

of 35 m and a roughness length of 0.0002 m, a correction factor of 0.896

would be needed to derive the equivalent 10-m winds.

7

The wind speeds recorded by ships are generally larger than those from

weather buoys and the corrected oil rig winds, but they become more

comparable when the ship observations are also corrected for height using the

assumption in the preceding paragraph (Fig. 6). As the height of the wind

measuring instruments can actually be different from the default height, and

the ships can be at a distance of up to 1 degree latitude/longitude away from

the fixed stations, the datasets are found to be less correlated and more

scattered. Yet, fitting the data using a straight line with zero intercept (y =

mx), slopes very close to one with biases less than 5% can be obtained. This

suggests that the raw wind reports from ships would generally over-estimate

the 10-m winds by about 10%. Caution should be taken in interpreting the

raw wind speed measurements from ships.

To test the sensitivity of the comparison with the choice of roughness

length, the above analysis are repeated using different values of the roughness

length as discussed in Section 3.2. The results are summarised in Table 5.

It could be seen that taking 0.0002 m as the roughness length would result in

the best matching among all wind observations datasets, with the wind speed

ratios between stations after height correction generally closest to one.

5. Wind profile for TC

Wind observations taken within the circulation of tropical cyclones are

relatively rare when compared with normal observations. Franklin et al.

(2003) studied the characteristics of tropical cyclone vertical wind profiles and

generated composite wind profiles using a large number of GPS

dropwindsondes. They confirmed the operational practices at the National

Hurricane Center (NHC) to adjust the flight-level winds to equivalent 10-m

winds as detailed in Franklin (2001). In essence, the surface wind reduction

factors for different heights have been derived from the dropwindsonde data

for the eyewall and outer vortex of TCs, and for the outer vortex region, the

reduction factors are further stratified by different quadrants and the

convective and non-convective regions of the TCs (Table 6). For ease of

comparison, the composite wind profiles and various reduction factors are

plotted alongside the log wind profiles under different roughness length values

in Fig. 7. It could be seen that, when compared with other roughness lengths,

the log wind profile with a roughness length of 0.0002 m is generally more

consistent with the composite profiles obtained from Franklin et al. (2003) for

8

heights below around 100 m. This further confirms the appropriateness of

using z0 = 0.0002 m for correcting the wind observations taken by the offshore

platforms for TC monitoring.

When a TC is tracking close to Hong Kong, ad-hoc surveillance flights

would be arranged jointly by HKO and GFS. In order to compare the wind

observations collected by the surveillance flights with those from the fixed

stations, pairing of flight observations with fixed station observations is first

conducted and the procedure is as follows: for each flight mission, a 6-hour

time window is manually chosen to include all observations within the window.

The relative positions of fixed stations or GFS aircraft to the TC centre at

observation times are then considered by referencing to the corresponding TC

location and movement, with the origin of the moving frame at any particular

time taken at the TC centre obtained by interpolation of the HKO best

track/operational warning dataset, and the positive y-axis pointing towards the

direction of movement of the TC. The relative distance and direction of a

fixed station or the GFS aircraft at different times will therefore vary with the

movement of the TC. The tangential and radial components of the wind

measurements from fixed stations and GFS aircraft are then averaged by

binning the relative distance of the observations at 10-km intervals, and the

binned wind components are paired up according to the relative distance to TC

centre. Considering the uncertainty in the TC centre fixes and the high wind

speed gradient near the eyewall, only relative distances between 100 km and

500 km are selected and assumed to be in the outer vortex region of the TC.

Besides, asymmetry of wind distribution usually exists in different quadrants

of the TC due to its movement and different reduction factors for adjusting

flight-level winds to surface level would apply. The binned dataset is further

stratified by quadrants dynamically chosen for more meaningful comparison.

To adjust the flight-level winds collected by surveillance flights, the

correction factors as given in Franklin (2001) are referenced. As the winds

measured could be taken within the convective or non-convective regions of

the TCs, correction factors averaged for both convective and non-convective

storms are used for simplicity. The adjustment is applied to the flight-level

winds according to the location of aircraft relative to TC motion.

Interpolated correction factors between heights are obtained based on

log-linear relationship and the factors for winds measured at 600 m over the

right/left/other (viz. front and rear) quadrants are 0.73/0.80/0.76, while the

9

factors for winds measured at 2500 m are 0.71/0.86/0.81. For most of the

cases in this study, the TCs concerned were located to the south of Hong Kong

and moved generally westward. A large portion of the sample data were

taken over the right semi-circle of the TCs.

Paired equivalent 10-m winds from fixed stations and surveillance flights

for seven TC cases during 2012-2014 are plotted in Fig. 8. Roughly

proportional trend is observed although the data points are more scattered as

the wind observations were not taken at the same location and time. For both

flight levels of 600 m and 2500 m, the adjusted wind observations from

aircraft and fixed stations are generally consistent with each other, with the

respective mean wind ratios both close to unity. The comparison suggests

that the reduction factors given in Franklin (2001) are reasonable for

operational use to adjust the flight-level winds conducted by GFS flights.

6. Discussion and Conclusions

In the operational analysis of the key parameters of a TC such as intensity

and various wind radii (strong, gale, and storm force winds), all available

observations over the ocean are ground-truth information and should be taken

into consideration. Performing height correction on the observations taken at

non-standard altitudes is the essential first step in order for an accurate analysis

to be made possible. This study has reviewed the interpretation of wind

observations from various offshore platforms for TC monitoring, with the

focus put on the method to correct the observations taken at different altitudes.

For wind measurements taken at an elevation of around 100 m or below,

the log wind profile as recommended by WMO (2012) is considered suitable

for operational deployment to retrieve the equivalent 10-m winds on open seas

neighbouring to Hong Kong. While the roughness lengths applicable to open

sea conditions range from 0.0002 m to 0.006 m in different implementations,

the lower bound of 0.0002 m appears to be the more appropriate choice for the

offshore platform data studied in this paper. This conclusion is supported by

the inter-comparison results between the observations taken by weather buoys,

ships, oil rig platforms, and co-located ASCAT winds, with high correlations

among the datasets and biases of correction generally within 5%.

The reduction to 10 m wind thus needed for a measurement taken at 100 m

10

above sea surface is as much as 17%, which highlights the necessity to correct

wind speeds obtained from elevated offshore platforms for the purposes of TC

analysis. For ship reports, the study reveals that the raw wind measurements

from ships generally over-estimate the 10-m winds by about 10%. In the

absence of detailed information about the method and height of observations to

allow for a more accurate correction to be made, caution should be exercised

to interpret the raw wind speed measurements from ships.

For GFS aircraft observations, wind correction factors from Franklin

(2001) for adjusting flight-level winds taken in the outer vortex of TCs have

been tested. Based on seven TC cases between 2012 and 2014, the corrected

wind measurements from GFS aircraft are generally consistent with the

equivalent 10-m winds derived from other offshore platforms. This

demonstrates the useful operational value of the aircraft observations for TC

monitoring.

Acknowledgements

The authors gratefully acknowledge the Guangdong Meteorology Bureau

of the China Meteorological Administration for providing the wind

observation data from weather buoys and oil rig platforms used in this study.

11

References

[1] Chan P.W., K.K. Hon, and S. Foster, 2011: Wind data collected by a

fixed-wing aircraft in the vicinity of a tropical cyclone over the south

China coastal waters. Meteorologische Zeitschrift, 20, 313-321.

[2] Franklin, J. L., 2001: Guidance for reduction of flight-level observations

and interpretation of GPS dropwindsonde data. NHC internal document.

[3] Franklin, J.L., M.L. Black, and K. Valde, 2003: GPS Dropwindsonde

Wind Profiles in Hurricanes and Their Operational Implications. Wea.

Forecasting, 18, 32-44.

[4] Giammanco, L.M., J.L. Schroeder, and M.D. Powell, 2013: GPS

Dropwindsonde and WSR-88D observations of tropical cyclone vertical

wind profiles and their characteristics. Wea. Forecasting, 28, 77-99.

[5] Harper, B.A., J.D. Kepert, and J.D. Ginger, 2010: Guidelines for

converting between various wind averaging periods in tropical cyclone

conditions. WMO TD-No.1555, 54 pp.

[6] Ingleby, B., 2009: Factors affecting ship and buoy data quality. Met

Office Meteorology Research and Development Technical Report No.

529, 37 pp.

[7] Kerith, F., S. Krumdieck, and J.F. Kreider, 2010: Principles of sustainable

energy. CRC Press, 895 pp.

[8] Thomas, B.R., E.C. Kent, and V.R. Swail, 2005: Methods to homogenize

wind speeds from ships and buoys. International Journey of Climatology, 25, 979–995.

[9] Verhoef, A., and A. Stoffelen, 2009: Validation of ASCAT 12.5-km winds,

version 1.2. Ocean and Sea Ice SAF Tech. Note

SAF/OSI/CDOP/KNMI/TEC/RP/147, 11 pp. [Available online at

http://www.knmi.nl/publications/fulltexts/validation_of_ascat_12.5km_winds_1.2.pdf.]

[10] WMO, 2012: Guide to meteorological instruments and methods of

observation WMO-No. 8, 716 pp.

12

Table 1 Roughness length for difference terrains as recommended by WMO (2012).

Terrain description Roughness length z0

(m)

Open sea, fetch at least 5 km 0.0002

Mud flats, snow; no vegetation, no obstacles 0.005

Open flat terrain; grass, few isolated obstacles 0.03

Crops, bushes, parkland 0.10 – 0.5

Regular large obstacle coverage (suburb, forest) 1.0

City centre with high- and low-rise buildings ≥ 2

Table 2 Roughness length parameters as recommended by Harper et al. (2010).

Terrain type Terrain description Roughness length z0

(m)

Sea

Open sea conditions for all wind

speeds, exposed tidal flats, featureless

desert, and tarmac.

0.0002 - 0.005

Smooth

Featureless land with negligible

vegetation such as wide beaches and

cays, exposed reefs

0.005 - 0.03

Open

Nearshore water for winds > 30 ms-1,

level country with low grass, some

isolated trees, airport surrounds.

0.03 - 0.10

Roughly

Open

Low crops, few trees, occasional

bushes. 0.10 - 0.25

Rough Lightly wooded country, high crops,

centres of small towns. 0.25 - 0.5

Very Rough Mangrove forests, palm plantations,

metropolitan areas. 0.5 - 1.0

Closed Mature regular rainforests, inner city

buildings (CBD) 1.0 - 2.0

Skimming

Mixture of large high and low-rise

buildings, irregular large forests with

many clearings.

> 2.0

13

Table 3 Roughness length in use for various kinds of surface by Meteo-France.

Kind of surface Roughness length z0

(m)

Open water 0.002 - 0.006

Naked ground 0.005 - 0.020

Short grass (1 cm high) 0.001

Rough grass (10 cm high) 0.023

Meadow (0.5 m high) 0.05 - 0.07

Wheat field (1 m high) 0.10 - 0.16

Table 4 Wind correction factors for various roughness lengths and heights.

Height (m) Roughness length z0 (m)

0.0002 0.0016 0.002 0.005 0.006

2 1.175 1.226 1.233 1.269 1.277

5 1.068 1.086 1.089 1.100 1.103

10 1.000 1.000 1.000 1.000 1.000

20 0.940 0.927 0.925 0.916 0.915

30 0.908 0.888 0.886 0.874 0.871

35 0.896 0.875 0.872 0.859 0.856

40 0.886 0.863 0.860 0.846 0.843

50 0.871 0.844 0.841 0.825 0.822

60 0.858 0.830 0.826 0.809 0.805

70 0.848 0.818 0.814 0.796 0.792

80 0.839 0.808 0.804 0.785 0.781

90 0.831 0.799 0.795 0.776 0.771

100 0.825 0.791 0.787 0.767 0.763

150 0.800 0.763 0.759 0.737 0.733

200 0.783 0.745 0.740 0.717 0.712

300 0.761 0.720 0.715 0.691 0.686

400 0.746 0.703 0.698 0.673 0.668

500 0.734 0.691 0.685 0.660 0.655

600 0.725 0.681 0.675 0.650 0.644

700 0.718 0.673 0.667 0.641 0.636

800 0.712 0.666 0.660 0.634 0.629

900 0.706 0.660 0.654 0.628 0.622

1000 0.701 0.655 0.649 0.623 0.617

14

Table 5 10 m wind speed ratios between various platforms assuming different roughness

lengths.

Log wind

profile

z0=0.0002

Corrected wind speed ratio (Left/Top)

Panyu oil rig

(107m)

Lufeng oil rig

(82m)

Ship

(35m)

Buoy

(10m)

ASCAT

(10m)

Panyu oil rig - 1.00 0.97 - 1.03

Lufeng oil rig - 0.98 - 0.95

Ship - 1.00 -

Buoy

- 1.00

Log wind

profile

z0=0.0016

Panyu oil rig

(107m)

Lufeng oil rig

(82m)

Ship

(35m)

Buoy

(10m)

ASCAT

(10m)

Panyu oil rig - 1.00 0.95 - 0.98

Lufeng oil rig - 0.97 - 0.91

Ship - 0.97 -

Buoy - 1.00

Log wind

profile

z0=0.002

Panyu oil rig

(107m)

Lufeng oil rig

(82m)

Ship

(35m)

Buoy

(10m)

ASCAT

(10m)

Panyu oil rig - 1.00 0.95 - 0.98

Lufeng oil rig - 0.96 - 0.91

Ship - 0.97 -

Buoy - 1.00

Log wind

profile

z0=0.005

Panyu oil rig

(107m)

Lufeng oil rig

(82m)

Ship

(35m)

Buoy

(10m)

ASCAT

(10m)

Panyu oil rig - 1.00 0.94 - 0.95

Lufeng oil rig - 0.96 - 0.89

Ship - 0.96 -

Buoy - 1.00

Log wind

profile

z0=0.006

Panyu oil rig

(107m)

Lufeng oil rig

(82m)

Ship

(35m)

Buoy

(10m)

ASCAT

(10m)

Panyu oil rig - 1.00 0.94 - 0.95

Lufeng oil rig - 0.95 - 0.88

Ship - 0.95 -

Buoy - 1.00

15

Table 6 Wind correction factors for flight level winds based on Franklin (2001).

Correction method

Correction factor

305m 925hPa 850hPa 700hPa

Dropsonde (eyewall) 0.80 0.75 0.80 0.90

Dropsonde (outer vortex)

(Left quadrant, convective) 0.80 0.80 0.85 0.90

Dropsonde (outer vortex)

(Left quadrant, non-convective) 0.80 0.80 0.80 0.85

Dropsonde (outer vortex)

(Other quadrant, convective) 0.80 0.75 0.80 0.85

Dropsonde (outer vortex)

(Other quadrant, non-convective) 0.80 0.75 0.75 0.80

Dropsonde (outer vortex)

(Right quadrant, convective) 0.80 0.70 0.70 0.75

Dropsonde (outer vortex)

(Right quadrant, non-convective) 0.80 0.70 0.65 0.70

Table 7 Wind correction factors for adjusting flight level winds collected by GFS flight

Location relative to TC Correction factor

600m 2500m

Right quadrant 0.73 0.71

Left quadrant 0.80 0.86 Other quadrant 0.76 0.81

16

Figure 1 Location map of weather buoys and oil rigs.

Figure 2 Correction factors to 10 m winds from log wind profile under difference

roughness lengths.

z0=0.005 m (Harper

et al., 2010)

z0=0.0002 m (HKO;

WMO, 2012)

z0=0.0016 m

(TurboWin)

z0=0.002~0.006 m

(RSMC La Reunion, Meteo-France)

17

Figure 3 Scatter plot of wind speed observations from weather buoys and ASCAT. Data

period: January 2012 - September 2014.

(a) (b)

(c) (d)

Figure 4 Scatter plot of wind speed observations from ASCAT and (a) Lufeng oil rig, (b)

Panyu oil rig, (c) Lufeng oil rig after correction to 10 m, and (d) Panyu oil rig after

correction to 10 m. Data period: January 2011 – September 2014.

y = 1.00x

R2= 0.95

N = 381

y = 0.88x

R2= 0.87

N = 464

y = 1.06x

R2= 0.87

N = 464

y = 0.80x

R2= 0.89

N = 452

y = 0.98x

R2= 0.89

N = 452

18

Figure 5 Scatter plot of corrected wind speed observations from Lufeng and Panyu oil rigs.

Data period: January 2011 – September 2014.

(a)

(b) (c)

Figure 6 Scatter plot of wind speed observations from ships and (a) weather buoys,

(b) Lufeng oil rig after correction to 10 m, and (c) Panyu oil rig after correction to 10 m.

Data period: January 2012 – September 2014.

y = 1.00x

R2= 0.61

N = 16633

y = 1.00x

R2= 0.42

N = 164

y = 1.02x

R2= 0.45

N = 118

y = 1.03x

R2= 0.45

N = 78

19

10

100

1000

0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1

z(m

)

v10/v

Log wind profile, z0=0.0002 m (HKO; WMO, 2012)

Log wind profile, z0=0.005 m (Harper et al., 2010)

Log wind profile, z0=0.0016 m (TurboWin)

Log wind profile, z0=0.002-0.006 m (Meteo-France)

Eyewall

Outer Vortex

Eyewall

Outer vortex / Left quad / non-convective

Outer vortex / Left quad / convective

Outer vortex / other quad / non-convective

Outer vortex / other quad / convective

Outer vortex / Right quad / non-convective

Outer vortex / Right quad / convective

Figure 7 Plot of the composite vertical wind profiles of TC from Franklin (2003)*, various

correction factors to 10-m winds from Franklin (2001)** and log wind profiles with

different roughness lengths.

(a)

(b)

Figure 8 Scatter plot of corrected wind speed observations from fixed stations and

surveillance flights derived from flight level winds at (a) 600 m and (b) 2500 m.

y = 0.99x

R2= 0.13

N = 50

y = 1.04x

R2= 0.47

N = 32

*

**