CHAPTER 5

measurement of surface wind

5.1

General

5.1.1

Definitions

The following definitions are used in this chapter

(see Mazzarella, 1972, for more details).
Wind velocity is a three-dimensional vector quantity
with small-scale random fluctuations in space and
time superimposed upon a larger-scale organized
flow. It is considered in this form in relation to, for
example, airborne pollution and the landing of
aircraft. For the purpose of this Guide, however,
surface wind will be considered mainly as a twodimensional vector quantity specified by two
numbers representing direction and speed. The
extent to which wind is characterized by rapid fluctuations is referred to as gustiness, and single
fluctuations are called gusts.
Most users of wind data require the averaged horizontal wind, usually expressed in polar coordinates
as speed and direction. More and more applications
also require information on the variability or gustiness of the wind. For this purpose, three quantities
are used, namely the peak gust and the standard
deviations of wind speed and direction.
Averaged quantities are quantities (for example, horizontal wind speed) that are averaged over a period
of 10 to 60 min. This chapter deals mainly with
averages over 10 min intervals, as used for forecasting purposes. Climatological statistics usually
require averages over each entire hour, day and
night. Aeronautical applications often use shorter
averaging intervals (see Part II, Chapter 2).
Averaging periods shorter than a few minutes do
not
sufficiently smooth the usually occurring natural
turbulent fluctuations of wind; therefore, 1 min
averages should be described as long gusts.

measure long smooth gusts; fast response systems

may indicate sharp wave-front gusts with a short
duration.
For the definition of gust duration an ideal measuring chain is used, namely a single filter that takes a
running average over t0 seconds of the incoming
wind signal. Extremes detected behind such a filter
are defined as peak gusts with duration t0. Other
measuring systems with various filtering elements
are said to measure gusts with duration t0 when a
running average filter with integration time t0
would have produced an extreme with the same
height (see Beljaars, 1987; WMO, 1987 for further
discussion).
Standard deviation is:

su = (ui U )2 =

(( ( u ) ( u ) n ) n )
i

(5.1)

where u is a time-dependent signal (for example,

horizontal wind speed) with average U and an overbar indicates time-averaging over n samples ui. The
standard deviation is used to characterize the
magnitude of the fluctuations in a particular signal.
Time-constant (of a first-order system) is the time
required for a device to detect and indicate about
63 per cent of a step-function change.
Response length is approximately the passage of
wind (in metres) required for the output of a windspeed sensor to indicate about 63 per cent of a
step-function change of the input speed.
Critical damping (of a sensor such as a wind vane,
having a response best described by a second-order
differential equation) is the value of damping
which gives the most rapid transient response to a
step change without overshoot.

Peak gust is the maximum observed wind speed

over a specified time interval. With hourly weather
reports, the peak gust refers to the wind extreme in
the last full hour.

Damping ratio is the ratio of the actual damping to

the critical damping.

Gust duration is a measure of the duration of the

observed peak gust. The duration is determined by
the response of the measuring system. Slowly
responding systems smear out the extremes and

Undamped natural wavelength is the passage of wind

that would be required by a vane to go through one
period of an oscillation if there were no damping. It
is less than the actual damped wavelength by a
factor (1 D 2 ) if D is the damping ratio.

I.52

Part I. Measurement of meteorological VARIABLES

Variable wind with no mean wind direction is wind where

the total variation from the mean wind direction
during the previous 10 minutes is 60 or more, and
less than 180, and the wind speed is less than 6 km/h
(3 kt), or when the total variation is 180 or more.
5.1.2

Units and scales

Wind speed should be reported to a resolution of

0.5 m s1 or in knots (0.515 m s1) to the nearest
unit, and should represent, for synoptic reports, an
average over 10 min. Averages over a shorter period
are necessary for certain aeronautical purposes (see
Part II, Chapter 2).
Wind direction should be reported in degrees to the
nearest 10, using a 01 ... 36 code (for example,
code 2 means that the wind direction is between 15
and 25), and should represent an average over
10min (see Part II, Chapter 2). Wind direction is
defined as the direction from which the wind
blows, and is measured clockwise from geographical north, namely, true north.
Calm should be reported when the average wind
speed is less than 1 kt. The direction in this case is
coded as 00.
Wind direction at stations within 1 of the North
Pole or 1 of the South Pole should be measured so
that the azimuth ring should be aligned with its
zero coinciding with the Greenwich 0 meridian.
There are important differences compared to the
synoptic requirement for measuring and reporting
wind speed and direction for aeronautical purposes
at aerodromes for aircraft take-off and landing (see
Part II, Chapter 2). Wind direction should be measured, namely, from the azimuth setting, with
respect to true north at all meteorological observing
stations. At aerodromes the wind direction must be
indicated and reported with respect to magnetic
north for aeronautical observations and with an
averaging time of 2 min. Where the wind measurements at aerodromes are disseminated beyond the
aerodrome as synoptic reports, the direction must
be referenced to true north and have an averaging
time of 10 min.
5.1.3

Meteorological requirements

Wind observations or measurements are required for

weather monitoring and forecasting, for wind-load
climatology, for probability of wind damage and
estimation of wind energy, and as part of the
estimation of surface fluxes, for example, evaporation
for air pollution dispersion and agricultural

applications. Performance requirements are given in

Part I, Chapter 1, Annex 1.D. An accuracy for
horizontal speed of 0.5 m s1 below 5 m s1 and better
than 10 per cent above 5 m s1 is usually sufficient.
Wind direction should be measured with an accuracy
of 5. Apart from mean wind speed and direction,
many applications require standard deviations and
extremes (see section 5.8.2). The required accuracy is
easily obtained with modern instrumentation. The
most difficult aspect of wind measurement is the
exposure of the anemometer. Since it is nearly
impossible to find a location where the wind speed is
representative of a large area, it is recommended that
estimates of exposure errors be made (see section 5.9).
Many applications require information about the
gustiness of the wind. Such applications provide
nowcasts for aircraft take-off and landing, windload climatology, air pollution dispersion problems
and exposure correction. Two variables are suitable
for routine reading, namely the standard deviation
of wind speed and direction and the 3 s peak gust
(see Recommendations 3 and 4 (CIMO-X) (WMO,
1990)).
5.1.4

Methods of measurement and

observation

Surface wind is usually measured by a wind vane

and cup or propeller anemometer. When the instrumentation is temporarily out of operation or when
it is not provided, the direction and force of the
wind may be estimated subjectively (Tables 5.1 and
5.2 provides wind speed equivalents in common
use for estimations).
The instruments and techniques specifically
discussed here are only a few of the more convenient
ones available and do not comprise a complete list.
The references and further reading at the end of this
chapter provide a good literature on this subject.
The sensors briefly described below are cup-rotor
and propeller anemometers, and direction vanes.
Cup and vane, propeller and vane, and propellers
alone are common combinations. Other classic
sensors, such as the pitot tube, are less used now for
routine measurements but can perform satisfactorily, while new types being developed or currently
in use as research tools may become practical for
routine measurement with advanced technology.
For nearly all applications, it is necessary to measure
the averages of wind speed and direction. Many
applications also need gustiness data. A wind-measuring system, therefore, consists not only of a sensor,
but also of a processing and recording system. The

Chapter 5. MEASUREMENT OF surface wind

I.53

Table 5.1. Wind speed equivalents

Beaufort scale number
and description

Wind speed equivalent at a standard height

of 10 m above open flat ground

Specifications for estimating speed over

land

(kt)

(m s1)

(km h1)

(mi h1)

<1

0 0.2

<1

<1

Calm; smoke rises vertically

Calm

Light air

13

0.3 1.5

15

13

Direction of wind shown by smoke-drift

but not by wind vanes

Light breeze

46

1.6 3.3

6 11

47

Wind felt on face; leaves rustle; ordinary

vanes moved by wind

Gentle breeze

7 10

3.4 5.4

12 19

8 12

Leaves and small twigs in constant

motion; wind extends light flag

Moderate breeze

11 16

5.5 7.9

20 28

13 18

Raises dust and loose paper; small

branches are moved

Fresh breeze

17 21

8.0 10.7

29 38

19 24

Small trees in leaf begin to sway, crested

wavelets form on inland waters

Strong breeze

22 27

10.8 13.8

39 49

25 31

Large branches in motion; whistling

heard in telegraph wires; umbrellas used
with difficulty

Near gale

28 33

13.9 17.1

50 61

32 38

Whole trees in motion; inconvenience felt

when walking against the wind

Gale

34 40

17.2 20.7

62 74

39 46

Breaks twigs off trees; generally impedes

progress

Strong gale

41 47

20.8 24.4

75 88

47 54

Slight structural damage occurs

(chimney-pots and slates removed)

10 Storm

48 55

24.5 28.4

89 102

11 Violent storm

56 63

28.5 32.6

103 117 64 72

12 Hurricane

64 and
over

32.7 and
over

118 and
over

processing takes care of the averaging and the

computation of the standard deviations and
extremes. In its simplest form, the processing can be
done by writing the wind signal with a pen recorder
and estimating the mean and extreme by reading
the record.

5.2

Estimation of wind

In the absence of equipment for measuring wind,

the observations must be made by estimation. The
errors in observations made in this way may be

55 63

Seldom experienced inland; trees

uprooted; considerable structural
damage occurs
Very rarely experienced; accompanied by
widespread damage

73 and
over

large, but, provided that the observations are used

with caution, the method may be justified as
providing data that would otherwise not be available in any way. If either temporarily or permanently
the wind data of some stations are obtained by estimation instead of measurement, this fact should be
documented in station records made accessible to
data users.
5.2.1

Wind speed

Estimates are based on the effect of the wind on

movable objects. Almost anything which is
supported so that it is free to move under the

I.54

Part I. Measurement of meteorological VARIABLES

Table 5.2. Wind speed equivalents for arctic areas and areas where there is no vegetation
Beaufort scale number
and description

Wind speed equivalent at a standard height

of 10 m above open flat ground

Specifications for estimating speed for

arctic areas and areas where there is no
vegetation

(kt)

(m s1)

(km h1)

(mi h1)

<1

0 0.2

<1

<1

Calm

Light air

13

0.3 1.5

15

13

No noticeable wind. Smoke rises nearly

vertically.

Light breeze

46

1.6 3.3

6 11

47

Wind felt on face, leaves rustle.

Gentle breeze

7 10

3.4 5.4

12 19

8 12

Hair is disturbed, clothing flaps.

Moderate breeze

11 16

5.5 7.9

20 28

13 18

Dust and loose paper raised, hair

disarranged.

Fresh breeze

17 21

8.0 10.7

29 38

19 24

Force of wind felt on body. Limit of

agreeable wind on land.

Strong breeze

22 27

10.8 13.8

39 49

25 31

Some inconvenience in walking.

Near gale

28 33

13.9 17.1

50 61

32 38

Difficulty when walking against wind.

Gale

34 40

17.2 20.7

62 74

39 46

Difficulty with balance in walking.

Strong gale

41 47

20.8 24.4

75 88

47 54

Danger in being blown over.

10 Storm

48 55

24.5 28.4

89 102

11 Violent storm

56 63

28.5 32.6

103 117 64 72

12 Hurricane

64 and
over

32.7 and
over

55 63

118 and
over

influence of the wind can be used, but the descriptive

specifications given in the Beaufort scale of wind
force, as reproduced in the table, will be found
especially useful.
In order to make the estimates, the observer (and the
wind-susceptible object) must stand on flat open
terrain as far as possible from obstructions. It must
always be remembered that even small obstructions
cause serious changes in wind speed and deviations
in wind direction, especially at their lee side.

Trees uprooted, considerable structural

damage.

73 and
over

5.2.2

Wind direction

In the case of an absence of instruments, or when

the instrumental equipment is unserviceable, the
direction should be estimated by observing the
drift of smoke from an elevated chimney, the
movement of leaves, and so on, in an open situation, or a streamer or pennant fixed to a tall
flagstaff. In addition, the wind drogue at an
airport may be used when the wind speed is sufficient to move such a device.

Chapter 5. MEASUREMENT OF surface wind

Whichever of these aids is used, errors due to

perspective are liable to be made unless the
observer stands vertically below the indicator.
Care should be taken to guard against mistaking
local eddies caused by buildings, and the like, for
the general drift of the wind.
In an open location, the surface wind direction
can be estimated rather accurately by facing the
wind. The direction of the movement of clouds,
however low, should not be taken into account.
5.2.3

Wind fluctuations

No attempt should be made to estimate peak

gusts or standard deviations without proper
instruments and recording devices.

5.3

Simple instrumental methods

At stations where orthodox anemometers cannot

be installed it may be possible to provide some
very low-cost, simple instruments that help the
observer take measurements that are somewhat
more reliable than those obtained by unaided
estimation.
5.3.1

Wind speed

Simple hand-held anemometers, if they are used,

should be set up and read in accordance with the
makers instructions. The measurement should be
taken from a point well exposed to the wind,
and not in the lee of obstructions such as buildings, trees and hillocks. If this is not possible, the
measurement point should be a good distance
from obstructions, namely at least 10 times the
obstruction height and upwind or sideways by at
least twice the obstruction height.
5.3.2

Wind direction

Direction may be estimated from a vane (or

banner) mounted on a pole that has pointers
indicating the principal points of the compass.
The vane is observed from below, and wind direction may be estimated to the nearest of the
16 points of the compass. If the vane oscillates in
the wind, the wind direction must be estimated
as the average direction about which the oscillations occur.

5.4

I.55

Cup and propeller sensors

Cup and propeller anemometers are commonly

used to determine wind speed and consist of
two sub-assemblies: the rotor and the signal
generator. In well-designed systems, the angular velocity of the cup or propeller rotor is
directly proportional to the wind speed, or,
more precisely, in the case of the propeller
rotor, to the component of the wind speed
parallel to the axis of rotation. Also, in welldesigned anemometers, the calibration linearity
is independent of air density, has good zero and
range stability, and is easily reproduced in a
manufacturing process. Near the starting
threshold, say for wind speeds of less than
4ms 1, the calibration of cup anemometers can
deviate substantially from linearity, if the arm
connecting the cup to the rotation axis is much
longer than the diameter of the cup (Patterson,
1926).
The nature of the response of the cup and propeller-type wind-speed sensors to changes in wind
speed can be characterized by a response length,
the magnitude of which is directly proportional
to the moment of inertia of the rotor and, in
addition, depends on a number of geometric
factors (Busch and Kristensen, 1976; Coppin,
1982).
For almost all cup and propeller-type wind
sensors, the response is faster for acceleration
than for deceleration, so that the average
speed of these rotors overestimates the actual
average wind speed. Moreover, vertical velocity fluctuations can cause overspeeding of cup
anemometers as a result of reduced cup interference in oblique flow (MacCready, 1966).
The total overspeeding can be as much as
10 per cent for some designs and turbulent
wind conditions (cup anemometers at 10 m
height with a response length of 5 m over very
rough terrain; Coppin, 1982). This effect can be
minimized by choosing fast-response anemometers, either cup anemometers of a design
verified as having a good cosine response or
propeller vanes that have virtually no vertical
component of overspeeding. In case that
performance cannot be investigated in a wind
tunnel, operational anemometers can be
compared in the field with a calibrated anemometer (Albers, Klug and Westermann, 2000).

I.56

Part I. Measurement of meteorological VARIABLES

Since both cup and propeller rotors turn with an

angular velocity that is directly proportional to
speed or to the axial component, they are particularly convenient for driving a wide variety of signal
generators. Alternating and direct current generators, optical and magnetic pulse generators, and
turn-counting dials and registers have been used
(WMO, 2001). The choice of signal generator or
transducer depends largely on the type of data
processor and read-out to be used. Care should be
taken to ensure that the bearings and signal generator have low starting and running frictional torques,
and that the moment of inertia of the signal generator does not reduce the response too much. In
cases of long-distance transmission, voltage signals
decrease due to cable resistance losses and are therefore inferior to pulse frequency signals, which are
not so affected during transmission.

synchros, digital angle-encoder discs, direct reading

dials and rotary switches have been used to advantage. The choice of signal generator is largely a
matter of the type of data processor and read-out
used. Care should be taken to ensure that the bearings and signal generator have low starting and
running frictional torques. The simplest recording
method is to have a sheet mounted around a cylinder rotating with the vane axis, on which a writing
instrument slowly travels downward.

The required and achievable characteristics for

wind-speed sensors are included in Part I, Chapter1,
Annex 1.D.

5.6

5.5

Wind-direction vanes

For the purpose of obtaining a satisfactory measurement, a wind vane will be suitable if it is well
balanced so as not to have a preferred position in
case the axis is not vertical. Multiple vane fins
should preferably be parallel to the vane axis,
because a vane with two fins at angles > 10 to its
axis has two equilibrium positions which each
differ significantly from the real wind direction
(Wieringa and van Lindert, 1971).
The response of the usual underdamped wind vane
to a sudden change in wind direction is normally
characterized by overshoot and oscillation about its
true position, with the amplitude decreasing
approximately exponentially. Two variables are
used to define this response: the undamped natural
frequency or wavelength and the damping
ratio, the ratio of the actual damping to the critical
damping (MacCready, 1966; Mazzarella, 1972). A
damping ratio between 0.3 and 0.7 is considered to
be good and as having not too much overshoot,
and a reasonably fast response (Wieringa, 1967).
Where a relatively long period average is to be
computed from data captured at short intervals, it
is self-evident that lower damping ratios may be
acceptable.
The signal generator is essentially a shaft-angle
transducer, and many varieties have been employed.
Potentiometers, alternating and direct current

The absolute accuracy of direction measurement

also depends on the care with which the instrument has been aligned to true north. The required
and achievable characteristics for wind-direction
vanes are included in Part I, Chapter 1, Annex 1.D.

Other wind sensors

Many physical principles can be used to measure

wind speed and direction, all of which have their
own merits and problems. New systems often have
been developed for specific purposes, such as smallscale fluctuations and air pollution studies (see, for
example, Smith (1980)). The following are other
types of sensors:
(a) Pitot tube anemometers, which measure the
overpressure in a tube that is kept aligned
with the wind vector by means of a direction
vane (see Gold (1936) and WMO (1984a) for
a description of the Dines anemometer). The
Dines linearizing recording system deals with
the speed averaging problem caused by the
quadratic relation between wind speed and
pressure, and it also provides useful gustiness
records without requiring electrical power;
(b) Sonic anemometers, which measure the
time between emission and reception of an
ultrasonic pulse travelling over a fixed distance
(Kaimal, 1980). Because sonic anemometers
have no moving parts owing to their principle, they have high durability and little accuracy deterioration;
(c) Hot-disc anemometers are recently developed
solid-state instruments which measure the
temperature gradient across a chip arrangement. This provides both wind speed and
direction at accuracies within the specification of Part I, Chapter 1, Annex 1.D (Van
Oudheusden and Huijsing, 1991; Makinwa,
Huijsing and Hagedoorn, 2001). They are
sturdy, and steady in calibration, but operational experience is limited so far;
(d) Hot-wire anemometers measure the cooling
of thin heated wires. Operationally they are

I.57

Chapter 5. MEASUREMENT OF surface wind

rather unreliable, both because of excessive

fragility and because their calibration
changes rather fast in unclean or wet
surroundings. They are not recommended
for use in precipitation;
(e) Antique swinging-plate vanes are a little
better than no instrument at all;
(f) Remote wind-sensing techniques with
sound (sodar), light (lidar) or electromagnetic waves (radar) are uncommon in
routine meteorological networks and will
not be discussed in this Guide. Details are
provided in Lenschow (1986).

5.7

Sensors and sensor

combinations for component
resolution

Propellers which respond only to the wind speed

component that is parallel to the axis of rotation of
the rotor can be mounted orthogonally to produce
two read-outs which are directly proportional to the
components in the axis directions. Other sensors,
such as twin-axis sonic anemometers, perform the
same function at the expense of more sophisticated
electronic adjuncts. Orthogonal propellers have the
disadvantage that exact cosine response (namely,
pure component sensitivity) is difficult to attain. A
cup anemometer/vane combination or a propeller
vane can also be used as a component device when
the velocity components are computed from the
measured wind speed and direction.

5.8

Data-processing methods

Signals from anemometer/vane combinations can

be processed and averaged in many different
ways. Before considering the aspects of the entire
wind-measuring chain (exposure, sensing, transmission, filtering, recording and processing), it is
useful to discuss the problem of averaging. This
Guide deals with the following outputs: averaged
horizontal wind (components or speed/direction),
standard deviations and peak gust.
5.8.1

Averaging

The averaging of wind vectors or their components is straightforward in principle, but there
are a few problems associated with it. The first is
that the mean vector speed in the average wind
direction U is less than the average of all instantaneous wind speeds by a small amount, generally

a few per cent (MacCready, 1966; Wieringa

1980a). If necessary, this may be corrected if the
standard deviation of wind direction sd is measured; for the ratio of U, and the averaged
instantaneous wind speeds is (Frenkiel, 1951):

(u

+ vi 2 = 1 sd 2 2

(5.2)

This effect of crosswind turbulence is often confused

with the overestimation (overspeeding), causing
distortion in the standard deviation s u (see
section 5.4).
The second problem is the discontinuity of the
wind direction between 0 and 360. This problem
can be solved either by recording on a cylinder or
by extending the recorder range (for example to
540 with an automatic device switching the
range from 0 to 360 and from 540 to 180), or by a
computer algorithm that makes successive
samples continuous by adding or subtracting
360 when necessary. The fact that the first-order
response of a cup anemometer and the secondorder response of a vane cannot be fully matched
is a problem of minor importance, because the
response differences are reflected only in the
high-frequency part of the fluctuations.
From the fundamental point of view, component
averaging is preferable over the independent
averaging of speed and direction. However, the
differences are very small and, for most
applications, component averages can easily be
derived from average speed and direction. This
also applies to the corresponding standard
deviations. From the technical point of view, the
independent treatment of speed and direction is
preferable for a number of reasons. First of all, the
processing of the signal for speed and direction is
independent, which implies that the operation of
one instrument can continue even when the other
drops out. Secondly, this data reduction is simpler
than in those cases where components have to be
computed. Lastly, the independent treatment of
speed and direction is compatible with common
usage (including SYNOP and SHIP coding).
The averages of horizontal wind speed can be
obtained with a number of both mechanical and
electrical devices. Perhaps the simplest example is a
mechanical rotation-counting register on a cup
anemometer commonly used to measure the
passage of wind during a chosen averaging time
interval. At the other end of the complexity spectrum, electrical pulse generators drive specialpurpose digital processors, which can easily

I.58

Part I. Measurement of meteorological VARIABLES

calculate averages, peak gusts and standard

deviations.
If wind speed and direction are recorded as continuous graphs, an observer can estimate 10 min
averages fairly accurately from a pen recording. The
recorded wind trace can also be used to read peak
gusts. The reading of dials or meters gives a feel for
the wind speed and its variability, but is subject to
large errors when averages are needed. Instantaneous
read-outs are, therefore, less suitable to obtain
10 min averages for standard weather reports.
5.8.2

Peak gusts and standard

deviations

The computation or recording of wind fluctuations is extremely sensitive to the dynamic

response of all the elements of the measuring
chain, including response length and damping
ratio of the sensors. Additionally, the dynamic
response of the system as a whole determines the
duration of peak gusts, as defined in section 5.1.1.
Slowly responding systems spread out the extremes
and indicate wide gusts with small amplitude,
whereas fast-response systems record high and
narrow peaks (gusts of short duration). It is clear
that the dynamic response of wind systems has to
be carefully designed to obtain gusts or standard
deviations that are accurate, reliable and compatible between stations.
Before specifying the appropriate response characteristics of wind-measuring systems, it is necessary
to define the gust duration as required by the
application. Wind extremes are mainly used for
warning purposes and for the climatology of
extreme loads on buildings, constructions and
aircraft. It is important to realize that the shortest
gusts have neither the time nor the horizontal
extent to exert their full damaging effect on large
constructions. WMO (1987) concludes that a gust
duration of about 3 s accommodates most potential users. Gusts that persist for about 3 s correspond
to a wind run (duration multiplied by the average wind speed) of the order of 50 to 100 m in
strong wind conditions. This is sufficient to engulf
structures of ordinary suburban/urban size and to
expose them to the full load of a potentially
damaging gust.
The standard deviation of wind direction and wind
speed can easily be computed with microcomputerbased equipment by taking samples of the signals at
intervals of about 1 s. Sampling frequencies should
not be too great, because the sensor itself provides
smoothing over a multiple of its response distance

(Wieringa, 1980b). A sampling frequency of 0.25 Hz

is suitable in most cases, but depends on the
response distance of the sensor and the wind speed.
Part III, Chapter 2, includes a detailed discussion of
the theory of sampling sensor signals.
Simultaneous computation of the standard deviation of the horizontal wind speed over 10 min
together with the detection of gusts with a duration
of a few seconds gives interesting requirements for
electronic filters. The gusts are most critical with
regard to filtering, so in practice the system is optimized for them. Any low-pass filter used for the
detection of peak gusts measured by fast anemometers, smoothing over a few seconds, may reduce
the standard deviation by up to 10 per cent. This
can be corrected if the filtering variables in the
measuring chain are well documented. Often, in
practice, the reduction is less because the standard
deviation increases if the average wind speed shows
a positive or negative trend. Alternatively, the unfiltered signal can be recorded separately for the
purpose of measuring an unbiased standard deviation. In the next section, recommendations are
made for wind-measuring systems with exact values
for the filter variables.
In order to determine peak gusts accurately, it is
desirable to sample the filtered wind signal every
0.25 s (frequency 4 Hz). Lower sampling frequencies can be used, but it should be realized that the
estimate of the extreme will generally be lower as
the extreme in the filtered signal may occur between
samples.
Apart from the wind vane inertial damping, any
further filtering should be avoided for wind direction. This means that the standard deviation of
wind direction can be determined within 2 per cent
with most wind vanes.
Accurate computation of the standard deviation of
wind direction requires a minimum resolution of the
digitization process, which is often done on the shaft
of the vane by means of a digital encoder. A 7 bit resolution is quite sufficient here because then a 5unit
for the standard deviation can still be measured with
an accuracy of 1 per cent (WMO, 1987).
5.8.3

Recommendations for the design

of wind-measuring systems1

Wind-measuring systems can be designed in many

different ways; it is impossible to cover all design
1 Recommended by the Commission for Instruments and
Methods of Observation at its tenth session (1989).

Chapter 5. MEASUREMENT OF surface wind

options in this Guide. Two common examples are

given here, one with mainly analogue signal
treatment and the other with digital signal
processing (WMO, 1987).
The first system consists of an anemometer with
a response length of 5 m, a pulse generator that
generates pulses at a frequency proportional to
the rotation rate of the anemometer (preferably
several pulses per rotation), a counting device
that counts the pulses at intervals of 0.25 s, and a
microprocessor that computes averages and
standard deviation over 10 min intervals on the
basis of 0.25 s samples. The extreme has to be
determined from 3 s averages, namely, by averaging over the last 12 samples. This averaging has
to be done every 0.25 s (namely, overlapping 3 s
averages every 0.25 s). The wind direction is
measured with a vane that has an undamped
wavelength of 5 m, a damping ratio of 0.3, and a
7 bit digital encoder that is sampled every second.
Averages and standard deviations are computed
over 10 min intervals, where successive samples
are checked for continuity. If two successive
samples differ by more than 180, the difference
is decreased by adding or subtracting 360 from
the second sample. With response lengths of 5 m
for the anemometer and the wind vane (damping
ratio 0.3, undamped wavelength 10 m), the
standard deviations of wind speed and wind
direction are reduced by about 7 and 2 per cent,
respectively. The gust duration corresponding to
the entire measuring chain (as defined in section
5.1.1) is about 3 s.
The second system consists of an anemometer
with a response length of 5 m, a voltage generator
producing a voltage proportional to the rotation
rate of the anemometer, analogue-to-digital
conversion every second, and the digital processing of samples. The wind-direction part consists of
a vane with an undamped wavelength of 5 m and
a damping ratio of 0.3, followed by analogue-todigital conversion every second and digital
computation of averages and standard deviations.
To determine peak gusts the voltage is filtered with
a first-order filter with a time-constant of 1 s and
analogue-to-digital conversion every 0.25 s. With
regard to filtering, this system is slightly different
from the first one in that standard deviations of
wind speed and direction are filtered by 12 per
cent and 2 per cent, respectively, while again the
gust duration is about 3 s. This system can also be
operated with a pen recorder connected to the
analogue output instead of the analogue-to-digital
converter. Only averages and extremes can be read
now, and the gust duration is about 3 s, unless the

I.59

pen recorder responds more slowly than the firstorder filter.

The signal-processing procedure, as described
above, is in accordance with Recommendation 3
(CIMO-X) (WMO, 1990) and guarantees optimal
accuracy. The procedure, however, is fairly complicated and demanding as it involves overlapping
averages and a relatively high sampling frequency.
For many applications, it is quite acceptable to
reduce the sampling rate down to one sample
every 3 s, provided that the wind signal has been
averaged over 3s intervals (namely, non-overlapping averaging intervals). The resulting gust
duration is about 5 s and the reduction in standard
deviation is 12 per cent (Beljaars, 1987; WMO,
1987).

5.9

Exposure of wind instruments

5.9.1

General problems

Wind speed increases considerably with height,

particularly over rough terrain. For this reason, a
standard height of 10 m above open terrain is
specified for the exposure of wind instruments.
For wind direction, the corresponding shift over
such a height interval is relatively small and can
be ignored in surface wind measurements. An
optimum wind observation location is one
where the observed wind is representative of the
wind over an area of at least a few kilometres, or
can easily be corrected to make it
representative.
For terrain that is uneven, contains obstacles, or is
non-homogeneous in surface cover, both wind
speed and direction can be affected considerably.
Corrections are often possible, and the tools to
compute such corrections are becoming available.
To improve the applicability of wind data, essential
information to perform such corrections should be
transmitted to the users in addition to the direct
measurements.
5.9.2

Anemometers over land

The standard exposure of wind instruments over

level, open terrain is 10 m above the ground.
Open terrain is defined as an area where the
distance between the anemometer and any
obstruction is at least 10 times the height of the
obstruction. Wind measurements that are taken
in the direct wake of tree rows, buildings or any
other obstacle are of little value and contain little

I.510

Part I. Measurement of meteorological VARIABLES

information about the unperturbed wind. Since

wakes can easily extend downwind to 12 or
15 times the obstacle height, the requirement of
10 obstruction heights is an absolute minimum.
In practice, it is often difficult to find a good or
even acceptable location for a wind station. The
importance of optimizing the location can hardly
be overstressed; nonetheless, it is difficult to give
universal guidelines.
Two aspects are very important. First, the sensors
should be kept away from local obstructions as
much as possible. When wind measurements are
taken on the side of masts or towers rather than
at their top, the instruments should be placed on
booms with a length of at least three mast or
tower widths (Gill and others, 1967). When wind
instruments are placed on top of a building, they
should be raised at least one building width above
the top. Secondly, the local situation should be
well documented (Wieringa, 1983). There should
at least be a map of the station surroundings
within a radius of 2 km, documenting obstacle
and vegetation locations and height, terrain
elevation changes, and so forth. Changes in the
surroundings, such as the construction of buildings or growth of trees nearby, should be explicitly
recorded in station logbooks. Station instrumentation should be specified in detail.
Where standard exposure is unobtainable, the
anemometer may be installed at such a height
that its indications should not be too much
affected by local obstructions and represent as far
as possible how the wind at 10 m would be if
there were no obstructions in the vicinity. If the
terrain varies little with azimuth, this may be
effected by placing the anemometer at a height
exceeding 10 m by an amount depending on the
effective surface roughness length z 0 of the
surroundings (see the annex): about 13 m if z0 =
0.1 m, and about 19 m if z 0 = 0.5m. Wieringa
(1980b) shows that the strategy of anemometer
height increase does not work well if local
sheltering varies strongly with azimuth. Simple
calculation procedures now exist to determine
the effect of local topography (Walmsley and
others, 1990), and the climatology of the gustiness
records can be used to determine exposure
corrections in inhomogeneous surroundings
(Verkaik, 2000). Evans and Lee (1981) and
Grimmond and others (1998) discuss the problem
in urban areas (see also Part II, Chapter 11).
In freezing weather, special precautions must be
taken to keep the wind sensors free from sleet and
ice accumulations. In some localities it may be

desirable to provide some form of artificial heating for the exposed parts such as a thermostatically
controlled infrared radiator. Sleet and ice shields
have been designed for particular types of wind
equipment (see Curran and others, 1977).
5.9.3

Anemometers at sea

There is an increasing requirement for instrumental

measurements of wind over the sea, especially by
means of automatic unattended systems (see also
Part II, Chapter 4). This task presents special problems since the standard exposure height of 10 m
specified for land use cannot always be achieved in
a marine environment owing to the state of the sea
and/or tidal height variation. The obvious extrapolation of the exposure criteria for land sites leads to
the idea that, on moored buoys, the anemometer
should be mounted 10 m above the waterline of the
buoy. However, other sources of error are often
more significant than those arising from different
exposure heights (for a review, see WMO, 1981). On
fixed platforms and ships, it is of the utmost importance that wind sensors be exposed sufficiently
high above the platform and its superstructure to
avoid the often extensive influence of the platform
on the local wind structure. In general, it is never
safe to assume that a wind sensor is unaffected by
the platform structure, even if it is exposed at least
10 m above the height of the tallest obstruction on
the platform, unless the platform is relatively small.
WMO (1981) concludes that, at sea, good exposure
should have higher priority in obtaining accurate
and useful measurements than standardization of
the measurements at 10 m (WMO, 1989). Despite
careful siting, it is often impossible in practice to
avoid exposure errors. In order to allow height and
flow distortion corrections to be made, it is very
important to keep a record and detailed information about anemometer location and platform or
ship type (shape, dimension). If wind speed is measured at a height significantly greater than 10m
(namely, when the appropriate reduction factor
would be > 1.2), a reduction to the 10 m level
should be performed according to the procedures
recommended in the following paragraph, and
using the constant for open sea in the table of the
annex.
5.9.4

Exposure correction

Surface wind measurements without exposure

problems hardly exist. The requirement of open,
level terrain is difficult to meet, and most wind
stations over land are perturbed by topographic
effects or surface cover, or by both (WMO, 1987;
Wieringa, 1996).

Chapter 5. MEASUREMENT OF surface wind

It is clear that exposure errors pose problems to

users of wind data and often make the data useless.
This problem is particularly serious in numerical
forecast models where there is a tendency to analyse
the wind and pressure fields separately. Surface
winds, however, can be used for initialization only
if they are representative of a large area. This means
that errors due to local exposure and/or non-standard measurement height must be removed.
The correction of wind readings for local exposure
can be performed only with measurements of
reasonable quality at locations that are not too
rough (z0 0.5 m) and reasonably level. No attempt
should be made to correct measurements that have
hardly any relation to a regional average. For example, a wind station in a deep valley, where the flow
is dominated by katabatic effects, may be important for local forecasts, but cannot be used as a
regionally representative wind.
If U is the wind speed measured at height z, the
corrected wind speed Uc which would be indicated
locally at 10 m above terrain with roughness z0
follows from:

ln (10 z0u
ou ) ln ( 60 z0u
ou ) ln (10 zo0 )
U c = U CF CT

ln ( z z 0u ) ln (10 zou
0u ) ln ( 60 zo0 )

(5.3)

where CF is the flow distortion correction; CT is the

correction factor due to topographic effects z0u is
the effective roughness length of the terrain
upstream of the measurement station, and z0 is
roughness length in the application (for example, a
grid box value in a numerical forecast model). In
this expression, z, z0 and z0u are specified in metres.
The different correction terms represent the
following:
(a) Flow distortion: The correction factor CF
accounts for flow distortion by nearby
big objects. This is particularly important
for anemometers on buildings, ships, and
platforms at sea. The best way of finding CF
as a function of wind direction is by means
of model simulation in a wind tunnel
(Mollo-Christensen and Seesholtz, 1967).
Estimates based on potential flow around
simple configurations can also be applied
(Wyngaard, 1981; WMO, 1984b). For
measurements on top of a free-standing
mast, flow distortion is negligible (CF = 1).
(b) Topographic correction: This correction
accounts for terrain height effects around
the wind station. CT is the ratio of the
regionally averaged wind speed (averaged
over ridges and valleys at 10 m above local

I.511

terrain) and the wind speed measured at the

wind station. In the example of an isolated
hill with a station at the top of the hill, CT
should be less than 1 to correct for the speedup induced by the hill, to make the result
representative of the area rather than of the
hill top only. CT equals 1 for flat terrain. For
isolated hills and ridges, estimates of CT can
be made with the help of simple guidelines
(Taylor and Lee, 1984). In more complicated
topography, model computations are needed
on the basis of detailed height contour maps
of the terrain surrounding the wind stations
(Walmsley and others, 1990). Such computations are fairly complicated but need to be
done only once for a single station and lead
to a semi-permanent table of CT as a function of wind direction.
(c) Non-standard measurement height: This
effect is simply included in the Uc formula
by assuming a logarithmic profile combined
with the roughness length z0u of the
upstream terrain. For stations over sea, this
reduction to standard height can be important, but stability corrections are relatively
small there, justifying the logarithmic form
of the reduction.
(d) Roughness effects: Upstream roughness
effects as well as the effects of surface obstacles can be corrected by extrapolating the
wind speed logarithmic profile to a height
of 60 m with the station specific effective
roughness length z0u and by interpolating
back to 10 m with the roughness length z0
necessary for the application. The roughness length z0u should be representative of a
2 km fetch upwind of the wind station; the
value usually depends on wind direction.
The annex discusses how to estimate z0u.
If flow distortion and topography problems are
negligible or have been corrected, apply the (c) to
(d) exposure correction by formula 5.3 towards z =
10m and z0 = 0.03 m. Corrected wind speeds then
will be equivalent to those which would have been
measured at a local hypothetical wind station
conforming fully with WMO requirements (10 m
over open terrain). Wind speeds corrected in this
way are called potential wind speeds (WMO, 2001).
Two comments are appropriate here. First, the
extrapolation height of 60 m should not be seen as
a very firm value. Heights between 40 and 80 m
would have been acceptable; 60 m is about the
correct magnitude in relation to the 2 km fetch for
which z0u is representative and has proved to give
satisfactory results (Wieringa, 1986). Secondly,
stability-related changes in the wind profile cannot

I.512

Part I. Measurement of meteorological VARIABLES

be neglected over the height range from 10 to 60 m,

but the effect of stability is relatively small in the
present formulation because the stability corrections
in the transformations upwards and downwards
cancel out. A practical example of the application
of wind measurement correction in an operational
context is given in WMO (2000) and WMO (2001).
Although most of the exposure correction can be
directly applied to the measurements, both
unadjusted (Level I) data and adjusted (Level II)
data are to be disseminated.

change in sensor characteristics leading to a deterioration in wind data quality may occur as a
result of physical damage, an increase in bearing
friction from the ingress of dust, corrosion, or
degradation of the transduction process (for
example, a reduction in the output of a cup or
propeller generator as a result of brush wear).

5.10

Instruments should be inspected for physical

damage, by checking the zero of the anemometer
system by holding the cups or propeller, and by
checking vane orientation by holding it fixed in a
predetermined position or positions. Repairs to
the sensors are usually only practicable in a
workshop.

Calibration and maintenance

A fully reliable calibration of cup, propeller and

vane anemometers is possible only in a wind
tunnel; the performance of such instruments is
now well known and the manufacturers calibration can be relied upon for most purposes, when
the instrument is in good condition. Wind-tunnel
tests are useful for special projects or for type-testing new models.
In the field, anemometers are prone to deterioration and regular inspections are advisable. A

The inspection of analogue traces will show faults

as indicated by incorrect zero, stepped traces due
to friction, noise (which may be evident at low
wind speeds), low sensitivity (at low speeds), and
irregular or reduced variability of recorded wind.

System checks should regularly be carried out on

the electrical and electronic components of electrical recording or telemetering instruments. Zero
and range checks should be made on both the
speed and direction systems.

Chapter 5. MEASUREMENT OF surface wind

I.513

Annex
The effective roughness length
For the purpose of exposure correction, a roughness
length z0 that represents the terrain over 2 km of
upstream fetch is needed as a function of wind
direction. The quality of the roughness correction
is very much dependent on the accuracy of this
roughness length.
Over sea, the task is relatively simple because of the
uniform fetch. The so-called Charnock relation can
be applied. It expresses the sea surface roughness to
the friction velocity u* and the gravitational acceleration g by means of z0u = u*2/g, where is an
empirical constant approximately equal to 0.014.
The friction velocity relates to the neutral wind
profile by means of U(z) = (u*/ ) ln (z/z0u), where is
the Von Karman constant (0.4) and z is the measurement height. These two equations have to be solved
iteratively, which can be done by starting with z0u =
0.0001, computing u* from the log-profile, evaluating z0u again, and repeating this a few times.
The surface roughness length over land depends on
the surface cover and land use and is often difficult
to estimate. A subjective way of determining z0u is
by a visual survey of the terrain around the wind
station with the help of the table below, the validity
of which has been recently corroborated (Davenport
and others, 2000). Choosing wind direction sectors
of 30 up to a distance of 2 km is most convenient.
With very non-homogeneous fetch conditions, an
effective roughness should be determined by averaging ln (z0u) rather than z0u itself.
The best way of determining z0u is with the help of
about one year of climatology of the standard deviations. The standard deviations of wind speed and
wind direction are related to the upstream roughness over a few kilometres and can be used for an
objective estimate of z0u. Both the standard deviation of wind speed su and the standard deviation of
wind direction sd (in radians) can be employed by
means of the following formulae:

su /U = cu

[ln (z/z0u)]1

A.1

sd /U = cv

[ln (z/z0u)]1

A.2

described in section 5.8.3, the standard deviation of

wind speed is filtered by about 12 per cent, and that
of wind direction by about 2 per cent, which implies
that cu and cv reduce to 1.94 and 1.86, respectively. In
order to apply the above equations, it is necessary to
select strong wind cases (U > 4 m s1) and to average
u/U and/or over all available data per wind sector
class (30 wide) and per season (surface roughness
depends, for example, on tree foliage). The values of
z0u can now be determined with the above equations,
where comparison of the results from u and give
some idea of the accuracy obtained.
In cases where no standard deviation information
is available, but the maximum gust is determined
per wind speed averaging period (either 10 min or 1
h), the ratios of these maximum gusts to the averages in the same period (gust factors) can also be
used to determine z0u (Verkaik, 2000). Knowledge
of system dynamics, namely, the response length of
the sensor and the response time of the recording
chain, is required for this approach.
Terrain classification from Davenport (1960)
adapted by Wieringa (1980b) in terms
of aerodynamic roughness length z0
Class
index

Short terrain description

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

Low crops; occasional large obstacles, x/H

> 20

0.10

High crops; scattered obstacles,

15 < x/H < 20

0.25

Parkland, bushes; numerous obstacles, x/H

10

0.5

Regular large obstacle coverage (suburb,

forest)

1.0

City centre with high- and low-rise

buildings

Note: Here x is a typical upwind obstacle distance and H is the

height of the corresponding major obstacles. For more detailed and

where cu = 2.2 and cv = 1.9 and = 0.4 for unfiltered

measurements of su and sd. For the measuring systems

updated terrain class descriptions see Davenport and others (2000)

(see also Part II, Chapter 11, Table 11.2).

I.514

Part I. Measurement of meteorological VARIABLES

References and further reading

Ackermann, G.R., 1983: Means and standard

deviations of horizontal wind components.
Journal of Climate and Applied Meteorology, 22,
pp. 959961.
Albers, A., H. Klug and D. Westermann, 2000:
Outdoor Comparison of Cup Anemometers.
DEWI Magazin, No. 17, August 2000.
Beljaars, A.C.M., 1987: The influence of sampling
and filtering on measured wind gusts. Journal
of Atmospheric and Oceanic Technology, 4,
pp. 613626.
Busch, N.E. and L. Kristensen, 1976: Cup
anemometer overspeeding. Journal of Applied
Meteorology, 15, pp. 13281332.
Coppin, P.A., 1982: An examination of cup
anemometer overspeeding. Meteorologische
Rundschau, 35, pp. 111.
Curran, J.C., G.E. Peckham, D. Smith, A.S. Thom,
J.S.G. McCulloch and I.C. Strangeways, 1977:
Cairngorm summit automatic weather
station. Weather, 32, pp. 6063.
Davenport, A.G., 1960: Rationale for determining
design wind velocities. Journal of the Structural
Division, American Society of Civil Engineers,
86, pp. 3968.
Davenport, A.G., C.S.B. Grimmond, T.R. Oke
and J. Wieringa, 2000: Estimating the
roughness of cities and sheltered country.
Preprints of the Twelfth American Meteorological Society Conference on Applied
Climatology (Asheville, NC , United States),
pp. 9699.
Evans, R.A. and B.E. Lee, 1981: The problem of
anemometer exposure in urban areas: a windtunnel study. Meteorological Magazine, 110,
pp. 188189.
Frenkiel, F.N., 1951: Frequency distributions of
velocities in turbulent flow. Journal of
Meteorology, 8, pp. 316320.
Gill, G.C., L.E. Olsson, J. Sela and M. Suda, 1967:
Accuracy of wind measurements on towers or
stacks. Bulletin of the American Meteorological
Society, 48, pp. 665674.
Gold, E., 1936: Wind in Britain The Dines and
some notable records during the last 40 years.
Quarterly Journal of the Royal Meteorological
Society, 62, pp. 167206.
Grimmond, C.S.B., T.S. King, M. Roth and
T.R. Oke, 1998: Aerodynamic roughness of
urban areas derived from wind observations.
Boundary Layer Meteorology, 89, pp. 124.

Kaimal, J.C., 1980: Sonic anemometers. Air-sea

Interaction: Instr uments and Methods
(F. Dobson, L. Hasse and R. Davis, eds), Plenum
Press, New York, pp. 8196.
Lenschow, D.H. (ed.), 1986: Probing the Atmospheric
Boundary Layer. American Meteorological
Society, Boston.
MacCready, P.B., 1966: Mean wind speed measurements in turbulence. Journal of Applied
Meteorology, 5, pp. 219225.
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. 182193.
Makinwa, K.A.A., J.H. Huijsing and A. Hagedoorn,
2001: Industrial design of a solid-state wind
sensor. Proceedings of the First ISA/IEEE Conference,
Houston, November 2001, pp. 6871.
Mazzarella, D.A., 1972: An inventory of specifications for wind-measuring instruments. Bulletin
of the American Meteorological Society, 53,
pp. 860871.
Mollo-Christensen, E. and J.R. Seesholtz, 1967:
Wind tunnel measurements of the wind disturbance field of a model of the Buzzards Bay
Entrance Light Tower. Journal of Geophysical
Research, 72, pp. 35493556.
Patterson, J., 1926: The cup anemometer.
Transactions of the Royal Society of Canada, 20,
Series III, pp. 154.
Smith, S.D., 1980: Dynamic anemometers. Air-sea
Interaction: Instr uments and Methods
(F. Dobson, L. Hasse and R. Davis, eds.). Plenum
Press, New York, pp. 6580.
Taylor, P.A. and R.J. Lee, 1984: Simple guidelines for
estimating wind speed variations due to small
scale topographic features. Climatological
Bulletin, Canadian Meteorological and
Oceanographic Society, 18, pp. 322.
Van Oudheusden, B.W. and J.H. Huijsing, 1991:
Microelectronic thermal anemometer for the
measurement of surface wind. Journal of
Atmospheric and Oceanic Technology, 8,
pp. 374384.
Verkaik, J.W., 2000: Evaluation of two gustiness
models for exposure correction calculations.
Journal of Applied Meteorology, 39, pp. 16131626.
Walmsley, J.L., I.B. Troen, D.P. Lalas and P.J. Mason,
1990: Surface-layer flow in complex terrain:
Comparison of models and full-scale observations.
Boundary-Layer Meteorology, 52, pp. 259281.

Chapter 5. MEASUREMENT OF surface wind

Wieringa, J., 1967: Evaluation and design of wind

vanes. Journal of Applied Meteorology, 6,
pp. 11141122.
Wieringa, J., 1980a: A revaluation of the Kansas
mast influence on measurements of stress and
cup anemometer overspeeding. BoundaryLayer Meteorology, 18, pp. 411430.
Wieringa, J., 1980b: Representativeness of wind
observations at airports. Bulletin of the
American Meteorological Society, 61,
pp. 962971.
Wieringa, J., 1983: Description requirements for
assessment of non-ideal wind stations, for
example Aachen. Journal of Wind Engineering
and Industrial Aerodynamics, 11, pp. 121131.
Wieringa, J., 1986: Roughness-dependent
geographical interpolation of surface wind
speed averages. Quarterly Journal of the Royal
Meteorological Society, 112, pp. 867889.
Wieringa, J., 1996: Does representative wind
information exist? Journal of Wind Engineering
and Industrial Aerodynamics, 65, pp. 112.
Wieringa, J. and F.X.C.M. van Lindert, 1971:
Application limits of double-fin and coupled
wind vanes. Journal of Applied Meteorology, 10,
pp. 137145.
World Meteorological Organization, 1981: Review
of Reference Height for and Averaging Time of
S u r f a c e Wi n d M e a s u r e m e n t s a t S e a
(F.W. Dobson). Marine Meteorology and
Related Oceanographic Activities Report No.
3, Geneva.
World Meteorological Organization, 1984a:
Compendium of Lecture Notes for Training Class
IV Meteorological Personnel (B.J. Retallack).
Volume II Meteorology (second edition),
WMO-No. 266, Geneva.
World Meteorological Organization, 1984b:
Distortion of the wind field by the Cabauw
Meteorological Tower (H.R.A. Wessels). Papers
Presented at the WMO Technical Conference on
Instruments and Cost-effective Meteorological
Observations (TECEMO), Instruments and
Observing Methods Report No. 15, Geneva.

I.515

World Meteorological Organization, 1987: The

Measurement of Gustiness at Routine Wind
Stations: A Review (A.C.M. Beljaars). Instruments
and Observing Methods Report No. 31,
Geneva.
World Meteorological Organization, 1989: Wind
Measurements Reduction to a Standard Level
(R.J. Shearman and A.A. Zelenko). Marine
Meteorology and Related Oceanographic
Activities Report No. 22, WMO/TD-No 311,
Geneva.
World Meteorological Organization, 1990: Abridged
Final Report of the Tenth Session of the Commission
for Instruments and Methods of Observation.
WMO-No. 727, Geneva.
World Meteorological Organization, 1991:
Guidance on the Establishment of Algorithms for
Use in Synoptic Automatic Weather Stations:
Processing of Surface Wind Data (D. Painting).
Report of the CIMO Working Group on Surface
Measurements, Instruments and Observing
Methods Report No. 47, WMO/TD-No. 452,
Geneva.
World Meteorological Organization, 1995: Manual
on Codes. Volume I.1, WMO-No. 306, Geneva.
World Meteorological Organization, 2000: Wind
measurements: Potential wind speed derived
from wind speed fluctuations measurements,
and the representativity of wind stations
(J.P. van der Meulen). Papers Presented at the
WMO Technical Conference on Meteorological
and Environmental Instruments and Methods of
Observation (TECO-2000), Instruments and
Obser ving Methods Report No. 74,
WMO/TD-No. 1028, p. 72, Geneva.
World Meteorological Organization, 2001: Lecture
notes for training agricultural meteorological
personnel (second edition; J. Wieringa and
J. Lomas). WMO-No. 551, Geneva (sections
5.3.3 and 9.2.4).
Wyngaard, J.C., 1981: The effects of probe-induced
flow distortion on atmospheric turbulence
measurements. Journal of Applied Meteorology,
20, pp. 784794.