Master’s thesis
Meteorology
Wind profile assessment for wind power purposes
Analysis and comparison of mast and SODAR measurements
from 17 sites in Finland
Iida Sointu
2014
Supervisors: Mira Hulkkonen, M.Sc., M.A.
Examiners: Heikki Järvinen, Prof.
Jouni Räisänen, Dos.
UNIVERSITY OF HELSINKI
DEPARTMENT OF PHYSICS
P.O. Box 64 Gustaf Hällströmin katu 2
FI-00014 University of Helsinki
HELSINGIN YLIOPISTO – HELSINGFORS UNIVERSITET – UNIVERSITY OF HELSINKI
Tiedekunta/Osasto – Fakultet/Sektion – Faculty/Section
Faculty of Science
Laitos – Institution – Department
Department of Physics
Tekijä – Författare – Author
Iida Sointu
Työn nimi – Arbetets titel – Title
Wind profile assessment for wind power purposes
Oppiaine – Läroämne – Subject
Meteorology
Työn laji – Arbetets art – Level
Master's thesis
Aika – Datum – Month and year
Sivumäärä – Sidoantal – Number of pages
January 2014
91
Tiivistelmä – Referat – Abstract
Preliminary estimation of wind speed at the wind turbine hub height is critically important when
planning new wind farms. Wind turbine power output is proportional to the cube of wind speed which
means that even small uncertainties in wind speed estimation will greatly affect the estimated energy
yield. Wind resource estimation is usually based on wind measurements using meteorological masts
and SODAR (SOnic Detection And Ranging) instruments.
Modern wind turbine hub height typically ranges from 100 to 140 meters. Wind speeds measured with
a mast must often be extrapolated to turbine hub height, while SODAR measures a continuous wind
profile up to approximately 200 meters height. The goal of this study is to assess the uncertainty of
SODAR measurements and to analyse the vertical wind profile variability in Finnish conditions from a
wind power perspective.
Mast and SODAR data collected at 17 sites across Finland covering a total of 381 months of
measurements was available for this study. Both the amount and type of equipment at the sites as
well as temporal data coverage varied greatly among the sites. Both coastal and inland sites were
represented.
Mast and SODAR data were quality controlled using manual inspection. The main reason for this was
to identify periods with anemometer freezing, which results in erroneous wind data. Quality controlled
data was used to calculate various parameters utilized in the wind resource assessment such as
annual wind speed, wind shear represented by the power law exponent (alpha), atmospheric stability
category and turbulence intensity. SODAR uncertainty was studied using the difference of wind speed
measured by co-located mast and SODAR in relation to wind speed, humidity, alpha and turbulence
intensity. Wind profile variability was studied with emphasis on hub height wind speed extrapolation,
particularly in terms of profile shape and wind speed magnitude.
Based on the results, the most common stability categories are neutral and slightly stable, which are
related to relatively strong wind shear. Wind profiles were found to vary significantly with season.
Therefore wind speed extrapolation should not be performed based on seasonal data when aiming for
statistical representativeness of the wind resource. As a result of this study a statistical method to
account for the seasonality is presented.
SODAR was observed to overestimate horizontal wind speed in rainfall as well as in high wind speed
and weakly turbulent conditions. Due to these tendencies, conducting short SODAR measurement
campaigns in which the afore-mentioned conditions prevail is not recommended. As winters have
generally higher wind speeds in Finland, short wintertime SODAR campaigns are discouraged.
A verification measurement period for mast-SODAR intercomparison ought to be carried out before
using SODAR as a standalone instrument. This should be conducted close to a meteorological mast
and the period should include as much meteorological variability as possible. This would ensure some
level of certainty in SODAR measurements, especially in the absence of an acknowledged calibration
standard.
Avainsanat – Nyckelord – Keywords
boundary layer, remote sensing, sodar, stability, wind energy, wind profile
Säilytyspaikka – Förvaringställe – Where deposited
Kumpula Campus Library
Muita tietoja – Övriga uppgifter – Additional information
Contents
1 Introduction
1.1 Previous studies . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Goals and hypotheses . . . . . . . . . . . . . . . . . . . . . . . . .
2 Background
2.1 Wind . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Wind in boundary layer . . . . . . . . . . . . . . .
2.2.1 Turbulence . . . . . . . . . . . . . . . . . .
2.2.2 Hydrostatic stability . . . . . . . . . . . . .
2.2.3 Vertical wind profile . . . . . . . . . . . . .
2.3 Wind power . . . . . . . . . . . . . . . . . . . . . .
2.4 Wind measurements . . . . . . . . . . . . . . . . .
2.4.1 Anemometers . . . . . . . . . . . . . . . . .
2.4.2 SODAR . . . . . . . . . . . . . . . . . . . .
2.4.3 Comparing mast and SODAR measurements
2.5 Long-term correlation of measurement data . . . .
1
4
5
.
.
.
.
.
.
.
.
.
.
.
7
7
8
9
10
13
17
18
19
22
25
25
3 Measurements
3.1 Mast measurements . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 SODAR measurements . . . . . . . . . . . . . . . . . . . . . . . .
27
29
30
4 Methods
4.1 Data cleaning . . . . . . . . . . . . . . . . . . . . .
4.2 Data analysis . . . . . . . . . . . . . . . . . . . . .
4.2.1 Data selection . . . . . . . . . . . . . . . . .
4.2.2 Mast and SODAR measurement comparison
4.3 Terminology . . . . . . . . . . . . . . . . . . . . . .
4.3.1 Availability . . . . . . . . . . . . . . . . . .
4.3.2 Turbulence intensity . . . . . . . . . . . . .
4.3.3 Height levels for the calculation of alpha . .
4.3.4 Season categories . . . . . . . . . . . . . . .
32
32
33
33
34
34
34
34
35
35
i
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4.3.5
Regression analysis parameters . . . . . . . . . . . . . . .
5 Results
5.1 Data processing . . . . . . . . . . . . . . . . . . .
5.1.1 Data cleaning . . . . . . . . . . . . . . . .
5.1.2 Measurement data quality . . . . . . . . .
5.1.3 Data availability . . . . . . . . . . . . . .
5.2 Atmospheric stability conditions . . . . . . . . . .
5.2.1 Stability estimation methods . . . . . . . .
5.2.2 Stability statistics based on measurements
5.3 Evaluation of SODAR reliability . . . . . . . . . .
5.4 Wind speed vertical profile . . . . . . . . . . . . .
5.4.1 Stability dependency . . . . . . . . . . . .
5.4.2 Seasonal variability . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
35
37
37
37
38
41
43
43
44
46
49
49
50
6 Conclusions and discussion
55
7 Appendices
7.1 Mast-SODAR regression analysis, part 1 . . . . . . . . . . .
7.2 Mast-SODAR regression analysis, part 2 . . . . . . . . . . .
7.3 Stability estimation methods . . . . . . . . . . . . . . . . . .
7.4 Dependency of wind speed difference on SODAR wind speed
7.5 Dependency of wind speed difference on turbulence intensity
7.6 Dependency of wind speed difference on relative humidity . .
7.7 Dependency of wind shear on stability category . . . . . . .
7.8 Seasonal wind profiles . . . . . . . . . . . . . . . . . . . . .
59
60
66
68
69
71
73
74
77
Bibliography
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
86
Chapter 1
Introduction
Wind power is a form of renewable energy. It is produced by converting the kinetic
energy of wind into electric potential. Energy production by wind power produces
no greenhouse gas emissions and contributes to the process of reducing energy
production by fossil fuels. In March 2013 the Finnish Government approved the
new National Energy and Climate Strategy including an objective to increase
wind energy production in Finland to 9 TWh by the year 2025 while the previous
target for 2020 was 6 TWh (Long-Term Climate and Energy Strategy). By the
end of 2012 the built turbine capacity was 288 MW with annual production of 492
GWh (Wind Energy Statistics in Finland). Figure 1.1 shows the development of
wind power production in Finland from 1992 to June 2013.
An essential part in developing new wind farms is the assessment of local wind
conditions. An accurate estimation of the wind conditions reduces the risks associated with the project as even small errors in predicted wind speeds affect
drastically the estimated energy yield. Wind resource assessment is based on
wind measurements conducted at or in the immediate surroundings of the site
(Burton et al., 2008). A typical measurement setup is a meteorological mast
equipped with cup anemometers on multiple, usually three, height levels up to
a height of approximately 100 meters. However, 100-meter masts rarely reach
the planned hub height (height of the center of turbine rotor) and higher masts
are expensive to build. The tip of an industrial size wind turbine rotor blade
may sweep at 200 meters height or above. In order to estimate wind speed at
1
CHAPTER 1. INTRODUCTION
Figure 1.1: Yearly wind power capacity and production in Finland from 1992 to 2013
(Wind Energy Statistics in Finland).
hub height and wind shear (change in wind speed with height) in the rotor swept
area, measurement results need to be extrapolated to higher levels which increases
the uncertainty of the wind speed estimations. Additionally, mast measurements
themselves can also be distorted in certain environmental conditions. For example, sub-zero temperatures may cause errors in anemometer measurements
due to ice formation.
Supporting measurements are often conducted with a SODAR (SOnic Detection And Ranging) which is an acoustic ground based remote sensing instrument (Bradley, 2010). It uses acoustic waves to measure vertical wind profile
(the change of wind speed and direction with height) in the lower atmosphere.
SODAR is a relatively affordable and compact in size compared to meteorological
masts. SODAR is easier to deploy and relocate than meteorological masts and
no permits for a SODAR siting are needed. One of the most useful aspects of a
2
CHAPTER 1. INTRODUCTION
SODAR is its ability to measure continuous wind profile up to 200 meters, typically with 5-meter height intervals. This enables the wind measurements to reach
any hub height with good vertical resolution. However, the uncertainties related
to the measurement technology are not fully understood. Due to the novelty of
SODAR as measuring instrument for wind power applications and complexity
of the interpretation of SODAR measurements they do not yet have a generally
recognized standard of best practices, unlike mast measurements. Photographs
of AQSystem AQ500 Wind Finder SODAR are presented in Figure 1.2.
Figure 1.2: AQ500 Wind Finder trailer and three speakers inside antenna (AQSystem).
The trailer makes the SODAR easy to deploy.
A comprehensive measurement campaign in a wind farm development project
should comprise several SODAR measurement periods and at least one 12 month
mast measurement period. SODAR measurements from multiple locations are
used to reduce spatial uncertainties in wind field estimation. Mast provides a
location-specific annual wind speed time series. By correlating the two measurement types information on the wind regime can be obtained at several locations
within the site, at height levels of interest and in a long-term basis (based on
mast measurement period).
As stated previously the mast measurements must be extrapolated to the desired
hub height. For this reason understanding the annual and stability-dependent
3
CHAPTER 1. INTRODUCTION
variability in wind profile shape is crucial. Sometimes wind resource assessments
are based on less than a full year of measurements. In these cases any extrapolation uses the seasonal wind profile shape, which may or may not be representative
of the annual profile. This discrepancy should be corrected for.
Seasonal wind measurements can be extended to cover a longer period using
Measure-Correlate-Predict (MCP) method with a reference data set. This gives
the corrected long-term wind speeds for the measurement levels but preserves the
wind profile shape. The seasonal discrepancy in the wind profile shape should
therefore be corrected separately.
1.1
Previous studies
Studies regarding SODAR measurement reliability have been published (for example Crescenti (1997) and Sanz Rodrigo et al. (2013)). The comparisons between
mast and SODAR measurements are generally unbiased but exhibit scatter at
high wind speeds. It should be noted that SODAR has to be placed relatively far
from the reference mast in order to eliminate clutter effects by the mast. Bradley et al. (2012) accounted for this and concluded that SODAR measures with
as good accuracy as mast. A comprehensive study on SODAR calibration technique was published in 2005 (Bradley et al., 2005), which supported the claim
that SODAR accuracy is comparable to anemometry. Moore and Bailey (2011)
formulated best practices for SODAR measurements including aspects such as
calibration, siting and comparisons with mechanical anemometry. This work was
later a contribution to Clifton et al. (2013) who compiled the best practices to
guide the use of remote sensing in wind power applications.
Sources of uncertainty in SODAR performance have been identified. Heavy rainfall can reduce SODAR data quality and introduce negative bias in relation to
mast measurements (Lang and McKeogh, 2011). According to Scott et al. (2010)
and Lang and McKeogh (2011), SODAR turbulence measurements are unreliable
due to unknown causes.
Studies on SODAR performance in different terrain types can be found in lit4
CHAPTER 1. INTRODUCTION
erature (Crescenti, 1997; Lang and McKeogh, 2011). Studies regarding complex terrain often concentrate on complex orography rather than complex surface
roughness or varying stability (for example Bradley, 2008 and Bradley et al.,
2012). Orography differences introduce horizontal variability to wind speed while
varying terrain roughness has a more pronounced effect on vertical wind field
structure.
The variation of vertical wind profile has been studied in relation to e.g. atmospheric stability (Touma, 1977; Gualtieri and Secci, 2011, 2013) and time (Smith
et al., 2002; Firtin et al., 2011). Most commonly used wind profile extrapolation
methods are studied by Bañuelos-Ruedas et al., 2010, Gualtieri and Secci, 2011
and Gualtieri and Secci, 2012. The widely used wind profile power law is shown
to approximate only neutral atmospheric conditions with good accuracy as wind
profile is dependent on atmospheric stability. Several studies have concluded that
the application of the power law wind profile on wind profile extrapolations may
lead to potentially large errors if only one value for wind shear is used and should
be used with caution (Rehman and Al-Abbadi, 2005; Ray et al., 2006; Fox, 2011).
Lackner et al. (2010) propose a method for estimating the hub height wind speed
using short-term hub height data (e.g. SODAR data) and 12-month below-hubheight data. The method assumes that there is some systematic relationship
between lower level and upper level wind shear parameter. The method is acknowledged to be sensitive to pronounced seasonal variations in the wind shear
at time-scales longer than the short-term hub height measurement period.
1.2
Goals and hypotheses
The first goal of this study is to better understand the uncertainty in SODAR
measurements in Finnish conditions. The effectiveness of measurement postprocessing to increase data quality is studied. The causes of uncertainty are analyzed by classifying the measurement conditions into geographical (e.g. coastal,
inland) and meteorological (e.g. stability, turbulence) conditions. The reliability
of SODAR measurements is assessed by comparing the measurements to near-by
mast measurements in different environmental conditions.
5
CHAPTER 1. INTRODUCTION
The second goal is to improve the understanding of spatial as well as temporal
variation of vertical wind speed profile. This study addresses the factors which
affect the wind profile in Finnish conditions where the forest covered terrain and
seasonal thermodynamic variability are defining features in wind profile shape.
For this study a comprehensive amount of mast and SODAR measurement data is
acquired from various locations. A total of 381 months of measurement data are
available. The measurement data and the locations of the sites are confidential
due to agreements with the data providers.
The hypothesis of the study is that high wind speed conditions are expected
to cause background noise to the measurements and thus lower the SODAR
measurement quality. Additionally, well mixed atmosphere where temperature
fluctuations are absent will manifest as low quality in SODAR measurements.
Distance from coastline is expected to influence overall stability at the site. Season
and stability are anticipated to be the most determining factors on vertical wind
profile shape.
6
Chapter 2
Background
2.1
Wind
Wind in the broadest sense is movement of air. Winds are caused by pressure
differences originating from differential heating of the Earth’s surface. This applies in a variety of spatial and temporal scales. Synoptic winds are ultimately
caused by the temperature difference between tropics and middle latitudes, while
for example sea breeze is caused by temperature difference between ground and
water surfaces.
When horizontal temperature differences are present, the temperature gradient
between two locations causes relative positive buoyancy in the warmer area and
negative in the colder area. This causes vertical air circulation which is compensated by horizontal airflow between the two locations (due to air being practically an incompressible fluid in the horizontal direction). The horizontal part
of the circulation is generally known as wind. The vertical wind component can
also be significant (for example inside cumulonimbus clouds), but wind usually
refers to the horizontal component.
Since air behaves like a fluid, the dynamics which govern the movement of air (i.e.
wind) can be understood by applying the Navier-Stokes equation (N-S) (Holton,
2004). N-S is the equivalent of Newton’s second law of motion for Newtonian
fluids. It describes how the velocity field of the fluid behaves when forces act on
7
CHAPTER 2. BACKGROUND
it. Formally the N-S for incompressible fluid can be written as:
→
−
1
∂V
→
−
→
−
→
− X→
−
+ ( V · ∇) V = − ∇p + V∇2 V +
Fi
∂t
ρ
i
(2.1)
→
−
where V is velocity field, ρ fluid density, p pressure field, V air viscosity and Fi
are the body forces acting on the fluid. The equation states that the velocity field
of fluid is produced by the pressure gradient, viscous stress and body forces. In
the case of air in the atmosphere, viscous stress contains the dynamic viscosity
and the acting body forces are gravity and apparent forces due to rotation of the
Earth and the fluid itself (Coriolis and centrifugal forces).
The Navier-Stokes equation is very general and is usually simplified using scale
analysis and boundary conditions for any practical use.
2.2
Wind in boundary layer
The planetary boundary layer is defined as the portion of the atmosphere in which
the wind field is directly influenced by the surface of the Earth. The boundary
layer can be divided into two general domains: surface layer and Ekman layer.
Surface layer refers to the lowest 10 % of the boundary layer and is also sometimes
referred to as constant flux layer or logarithmic wind profile layer (due to the
shape of the wind profile) (Stull, 1988). In Finland the depth of the boundary
layer typically ranges from circa 100 meters up to well over one kilometer.
Wind in the boundary layer is created by synoptic scale pressure gradient which
introduces the background wind field. This wind field is then deformed and
perturbed by local conditions such as surface roughness, orography, atmospheric
stability and mesoscale dynamics (for example sea breeze) (Holton, 2004).
Boundary layer wind can be expressed using a simplified form of the N-S equation
(see Equation 2.1) called the Boussinesq approximation (Holton, 2004). The main
assumption in this approximation is that density fluctuations are important in
the buoyancy term of the momentum equations, but negligible in all other terms.
The approximation is as follows:
8
CHAPTER 2. BACKGROUND
Du
Dt
Dv
Dt
Dw
Dt
0
= − ρ10 ∂p
+ f v + Fx
∂x
0
= − ρ10 ∂p
− f u + Fy
∂y
=
0
− ρ10 ∂p
∂z
+
θ0
g
θ
(2.2)
+ Fz
where u,v and w are the three dimensional wind components, ρ0 is air density, p0 is
pressure fluctuation, f is the Coriolis parameter, g is gravitational acceleration, θ
and θ0 are potential temperature mean and fluctuation respectively and Fi are the
friction force components. Potential temperature can be seen as the measure of
heat energy within an air parcel. Potential temperature profile is used to describe
hydrostatic stability within the boundary layer.
In the horizontal direction, wind is produced by the horizontal pressure gradient
and consumed by the horizontal friction. In the vertical direction wind is produced
by both the vertical pressure fluctuation from hydrostatic balance and buoyancy
(signified by the potential temperature θ) caused by heating. Any net production
is again counteracted eventually by the friction term.
2.2.1
Turbulence
Wind in the atmosphere is susceptible to turbulence. Atmospheric turbulence
exhibits as chaotic and irregular vortices in the wind field. These vortices are
superimposed on the mean wind field and are, in the light of the current understanding, impossible to fully predict. For this reason turbulence in the atmosphere
is analyzed in a stochastic manner. There are several ways to quantify and analyze turbulence, but all rely on certain conditions and assumptions and often
involve experimental parameters.
The physical origins of turbulence in the atmosphere can be divided into two
general categories. Mechanical turbulence is inherent in the wind field itself and
is caused by the three dimensional change in wind speed in all spatial scales.
This change is usually referred to as wind shear. Wind shear causes vortices to
form in the wind field which in turn affect the wind field itself, causing wind to
be chaotic and unpredictable. Mechanical turbulence is most prevalent near the
9
CHAPTER 2. BACKGROUND
ground where vertical wind shear is strongest and the surface effects (orography,
different surface types) contribute to turbulence formation.
Thermal turbulence occurs during convection. In a convective fluid the updrafts
and counteracting downdrafts are structured as distinct cells. Due to reversal of
flow direction between the cells horizontal shear develops which in turn causes
turbulent eddies to form. As convection requires heating from the surface this
kind of turbulence is called thermal turbulence (Stull, 1988).
Turbulence intensity
In wind energy applications turbulence intensity (TI) is defined as:
σv
(2.3)
v̄
where σv is standard deviation of wind speed and v̄ is mean wind speed. Both
the standard deviation and mean are calculated from 10–60 minute averaged
observations. TI varies with height, surface roughness and stability. Typical
values for I over different terrains are: sea 8 %, flat open grassland 13 % and
complex terrain 20 % or more (Petersen et al., 1998).
I=
2.2.2
Hydrostatic stability
Hydrostatic stability (shortened to stability) depends on the vertical temperature profile of the atmosphere. Stability has a significant effect on the turbulent
mixing of several key properties in planetary boundary layer such as momentum,
heat energy and humidity. Different mixing conditions manifest as differing vertical profiles in many observable atmospheric properties, including wind speed,
temperature and water vapor mixing ratio.
Atmospheric stability can be classified into three general classes; stable, unstable
and neutral. In the neutral case, the boundary layer is ’well mixed’ in the sense
that the effect of the surface is freely propagated through the boundary layer.
The potential temperature profile is constant which means average vertical mass,
momentum and energy fluxes are nonexistent, yet momentum is exchanged in
10
CHAPTER 2. BACKGROUND
vertical dimension by turbulent flux. Using this assumption the logarithmic wind
profile law (see Equation 2.10) can be derived.
In an unstable atmosphere the potential temperature decreases in the vertical,
causing the boundary layer to be in a constant state of vertical mixing. Wind is
weaker throughout the boundary layer (except near the surface) compared to the
neutral state due to more efficient mixing of momentum from the atmosphere to
the ground. In other words, the decelerating effect of surface friction is more easily
transported into the upper altitudes. An unstable atmosphere will eventually
reach the neutral state when the vertical potential temperature gradient reaches
zero.
A stable atmosphere is marked by the increase of potential temperature in the
vertical and there is little to no buoyancy induced vertical mixing. This confines
the surface friction effect to a shallow layer near the ground. The layer above
experiences less surface friction due to the convective inhibition and winds in
the upper boundary layer are stronger than in either neutral or unstable states.
Different stability classes are presented schematically in Figure 2.1.
Richardson number
Richardson number is a measure of stability. In particular, it measures the importance of thermal turbulence relative to mechanical turbulence. One form of
Richardson number is bulk Richardson number Rib which can be calculated based
on wind and temperature gradients between two levels (1 and 2):
Rib =
θ2 − θ1
(z2 − z1 )
0.5(θ1 + θ2 ) (u2 − u1 )2
g
(2.4)
Stability classification
There are several ways to classify the atmospheric stability based on different
measurement parameters. Perhaps the most widely used is the Pasquill-Gifford
(P-G) classification which uses seven categories from A to G. It requires records
of horizontal wind speed, cloud cover, ceiling height and time of observation.
11
CHAPTER 2. BACKGROUND
Figure 2.1: The effect of stability on wind profile shape and vertical mixing. In panel
(e) wind profile is re-plotted with a logarithmic height scale (Oke, 1978).
Since cloud cover and ceiling measurement equipment are expensive there are
alternative methods to determine the P-G categories.
Indicators of atmospheric stability and turbulent mixing that other methods use
are e.g. standard deviation of horizontal wind direction, standard deviation of
elevation angle of the vertical wind direction, standard deviation of the vertical
wind speed, wind speed at 10 meter height and solar radiation (Atmospheric
stability classification, 2013). The methods to determine the P-G categories vary
in complexity. For this reason they are not only compromises between accuracy
and practicality (through available observations), but emphasize different aspects
of atmospheric stability.
A simple method to determine the P-G stability category is the sigma theta
method which uses observations of standard deviation of horizontal wind direction. Another simple method is the delta T method, which uses temperature
12
CHAPTER 2. BACKGROUND
measurements from two or more heights to determine the temperature lapse rate
which in turn is used to estimate stability. Since the delta T method uses no
wind information it describes thermal turbulence. The sigma theta method on
the other hand describes mostly mechanical turbulence. Table 2.1 relates sets of
ranges of the defining parameters in the sigma theta and temperature gradient
methods to the P-G categories.
Table 2.1: Ranges of the defining parameters in the sigma theta and delta T methods
in relation to the Pasquill-Gifford categories (nrc, 1980).
Stability classes
Extremely unstable
Moderately unstable
Slightly unstable
Neutral
Slightly stable
Moderately stable
Extremely stable
P-G category Sigma theta
Delta T
◦
◦
[]
[ C/ 100 m]
A
>22.5
<-1.9
B
17.5-22.5
-1.9–1.7
C
12.5-17.5
-1.7 to -1.5
D
7.5-12.5
-1.5 to -0.5
E
3.8-7.5
-0.5 to 1.5
F
2.1-3.8
1.5 to 4.0
G
<2.1
>4.0
It has been shown that more complex methods based on either the Richardson
number or the M-O length (see Equation 2.13) generally yield more accurate
results, since they take both mechanical and thermal turbulence into account.
They are however more difficult to apply and therefore care should be taken
when choosing the suitable stability classification method (Sanz Rodrigo et al.,
2013; Hunter, 2012).
2.2.3
Vertical wind profile
Vertical profile of horizontal wind (shortened to wind profile) refers to the horizontal wind vector as a function of height. Due to the surface influence wind
speed in boundary layer can change significantly from the surface values. The
exact shape of the wind profile depends on surface roughness and atmospheric
stability. Wind direction is approximately constant near the surface and begins
to change only in the Ekman layer above. For this reason wind profile near the
surface usually refers to simply wind speed as a function of height.
13
CHAPTER 2. BACKGROUND
The shape of the wind profile is important to quantify and express formally. In a
hydrostatically neutral atmosphere mean vertical transport of momentum is close
to zero. For this reason the vertical momentum balance is solely determined by
turbulent momentum transport. The momentum transport can be written as:
∂(u0 w0 )
,
(2.5)
∂z
where the argument of the derivative is the turbulent momentum flux and u0 and
w0 are the turbulent parts of the horizontal and vertical wind speeds, respectively
(Tamura et al., 2007).
Using a first order closure called the K-theory, the turbulent momentum flux can
be rewritten as:
∂u
(2.6)
u0 w0 = −K ,
∂z
where K is eddy viscosity and u is mean wind speed. The above equation is similar
to the diffusion equation which hints that the K-theory applies to the case where
turbulent mixing is caused by very small eddies. This is a valid assumption in
the neutral atmosphere where mixing is caused by wind shear turbulence instead
of buoyancy effects.
On the other hand, Prandtl mixing length theory gives the parametrization of K
in neutral surface layer as:
K = l2
∂u
∂u
= k2z 2
∂z
∂z
(2.7)
where l is the Prandtl mixing length and the constant k is a universal constant
called von Karman’s constant, which has an experimentally determined value of
k ≈ 0.4 (Holton, 2004). Additionally, in neutral surface layer the turbulent flux
is approximately constant in the vertical:
u0 w0 (z) ≈ u0 w0 (z = 0) = u2∗
where u2∗ is called friction velocity.
14
(2.8)
CHAPTER 2. BACKGROUND
Combining the above equations we obtain the following:
u0 w 0
=
u2∗
2 2
=k z
∂u
∂z
2
∂u
∂z
k
⇒ u∗ = kz
⇒
= ∂u
∂z
z
u∗
(2.9)
The above equation yields the logarithmic wind profile when integrated in respect
to height (z). The logarithmic wind profile is therefore:
u(z) =
u∗
z
ln
k
z0
(2.10)
where z0 , the roughness length, is a constant of integration chosen so that ū = 0
at z = z0 .
The underlying assumption in the logarithmic wind profile is that there are no
obstructions to the airflow above the ground. In a forested area this assumption
is not valid, and the profile must be adjusted accordingly. The logarithmic profile
only applies above the forest canopy and for this reason the height z is reduced
by what is called the displacement height d. This is illustrated in Figure 2.2 and
formulated as following:
u∗ z − d
u(z) =
ln
(2.11)
k
z0
The value of the displacement height is generally two thirds of the forest height.
In order to use the logarithmic wind profile law the surface roughness must be
known and atmospheric stability must be neutral. To estimate the wind profile
when these conditions are not fulfilled one may use the wind profile power law
(Peterson and Hennessey, 1978). The power law is formally:
u(z) = ur
z
zr
α
(2.12)
where ur is the reference wind speed at reference height zr , z is height above
ground and α is an empirically derived coefficient. The formula is simpler than the
logarithmic law since all the information of the surface roughness and atmospheric
stability are combined into a single parameter α. The power law exponent α (later
referred to as alpha) for wind speed varies with height and also seasonally with
15
CHAPTER 2. BACKGROUND
Figure 2.2: Wind profile in forested areas where zd in the figure denotes the displacement
height d (Junge and Westerhellweg, 2013).
highest values in winter. Typical range of alpha in Finland is from 0.2 up to 0.6.
If available measurement levels do not reach the hub height, vertical extrapolation
of wind speed is needed. Extrapolation can be performed by fitting the power
law profile to the wind speed measurements and using the resulting profile to
deduce the hub height wind speed. Extrapolation causes uncertainty in the energy
yield calculations which is taken into account by adding an uncertainty to the
estimated wind speed (e.g. 3 % of wind speed for 20 meter extrapolation) at the
hub height. The above mentioned profile laws are applicable in the surface layer
below approximately 200 m of height (Cenedese et al., 1998).
Monin-Obukhov similarity theory
The simplest forms of wind profile laws only apply in neutral stability. In case of
non-neutral stability the commonly used logarithmic and power law alpha profiles
16
CHAPTER 2. BACKGROUND
cannot be used. In these cases the formulas for the neutral stability must be
modified using Monin-Obukhov similarity theory (Foken, 2006). The logarithmic
profile law with stability correction is then:
u∗
z
z
u(z) =
ln
− ψm
k
z0
L
(2.13)
where L is the Monin-Obukhov length and ψm is stability function. In physical
terms L is the height where turbulent kinetic energy production by convection (or
buoyancy) is equal to that by mechanical turbulence. The stability function ψm is
negative in stable conditions (wind shear increases) and positive in unstable conditions (wind shear decreases). However, typical wind measurement campaigns
provide insufficient observations for determining the stability function. In order
to determine the Monin-Obukhov length L friction velocity and kinematic heat
flux much be calculated, which in turn require turbulence measurements.
2.3
Wind power
Wind turbines convert kinetic energy in wind to mechanical energy. The power
output of a wind turbine is given by:
1
P = Cp ρAu3
2
(2.14)
where Cp is the power coefficient, ρ is the density of air, A is the rotor swept area
and u is the wind speed (Burton et al., 2008). As can be seen from Equation
2.14 the energy yield of a wind turbine is relative to the cube of wind speed.
This means that even small fluctuations in wind speed will result in substantial
change in energy yield. The dimensions of the turbines also play a significant part
in energy yield. The greater the rotor diameter and the higher the hub height
(distance between ground and rotor center), the greater the yield. However,
large turbines are usually more expensive to build and experience more structural
stress.
17
CHAPTER 2. BACKGROUND
Both the hub height and rotor diameter of new wind turbines are continuously
increasing (see Figure 2.3). For example in Finland the recently deployed wind
turbines in Ii Olhava wind farm are 140 meters of hub height with a rotor diameter
of 112 meters (TuuliWatti Oy, 2013). As the typical height of boundary layer in
Finland is from circa 100 meters up to well over one kilometer, wind turbines are
normally within the boundary layer.
Figure 2.3: Wind turbine size increase from 1980 to 2011. The figure shows relative size
of the swept area as turbine power output increased from 75 kW to 7.5 MW (Delphi234,
2012).
2.4
Wind measurements
In wind power industry wind measurements are typically conducted with anemometers mounted on a meteorological mast constructed for this purpose. More
modern wind measurement technologies such as remote sensing using SODAR
equipment are becoming increasingly common due to their flexibility. Since remote sensing devices are not the traditional method for conducting wind measurements in wind power industry their performance needs to be assessed by comparison with standardized mast measurements. Photographs of mast installation
and AQ500 Wind Finder SODAR are presented in Figure 2.4.
Both anemometers and SODAR devices should be periodically calibrated to document their response to changes in the environmental conditions. All the meas18
CHAPTER 2. BACKGROUND
urement data used in wind resource assessment need to be quality controlled.
Figure 2.4: Photographs of measurement mast and SODAR. The telescopic lattice meteorological mast has three measurement levels. Measurements are conducted at both
sides of the mast with anemometers at the end of the horizontal booms. SODAR instrument has a cone on top of the receiver to protect it. Photographs are owned by
Pöyry Finland Oy.
2.4.1
Anemometers
Wind is commonly measured using two distinct types of anemometers: cup and
ultrasonic. Cup anemometers consist of a set of cup-like arms which rotate with
wind. The angular velocity of the arms is converted to wind speed. Since this
setup gives only wind speed the cup anemometer is commonly coupled with a
wind vane for measuring wind direction. The IEC 61400-12-1 standard for turbine power curve measurements (International Electrotechinal Commission, 2005)
specifies the reference wind measurement system as a mast equipped with calibrated mechanical cup anemometers. The responses of traditional cup anemometers to changes in wind speed and the sources of error are fairly well understood.
19
CHAPTER 2. BACKGROUND
Cup anemometers suffer from underspeeding in freezing conditions due to ice
build-up on the instrument. In Finland freezing conditions are common during
winter (Finnish Meteorological Institute (FMI) and Technical Research Centre of
Finland (VTT)).
Ultrasonic anemometers have three or more sensor arms equipped with a set of
transducers. The anemometer operates by measuring the speed of acoustic signals
between the arms and deducing the wind speed from the measurements. Since
the anemometer has three sensor arms, both wind speed and direction can be
obtained. A 3-D ultrasonic sensor has another pair of sensor arms on top of the
other enabling the measurement of vertical wind as well. Ultrasonic measurements
can be made with a frequency exceeding 20 Hz which enables the estimation of
wind turbulence in addition to the mean wind (Wyngaard, 1981). An example of
a cup anemometer and an ultrasonic anemometer are presented in Figure 2.5.
Figure 2.5: Thies First Class wind velocity sensor and Thies 2D ultrasonic anemometer
(Thies Clima).
Both measurement types have their pros and cons. Cup anemometer is inexpensive and widely utilized in meteorological measurement sites. Ideally, a cup
anemometer would respond to only the horizontal component of the wind vector.
However, in complex terrain or in very turbulent conditions cup anemometers
may also respond to the vertical wind. It has been shown (Kristensen, 1998;
Brock and Richardson, 2001; Moore and Bailey, 2006) that in turbulent air flow
20
CHAPTER 2. BACKGROUND
cup anemometers suffer under-speeding in a gust and over-speeding when the gust
has passed. These effects do not cancel out as the slowing down takes more time
and consequently the average wind speed is over-estimated. This is illustrated in
Figure 2.6.
Figure 2.6: Response of a cup anemometer to a simple wind speed input (Brock and
Richardson, 2001).
Ultrasonic anemometers provide additional information from the wind field but
are more expensive, consume more power and are technologically more complex
than mechanical cup anemometers. In addition to the mean wind field they
can estimate the turbulent components of the wind due to the high sampling frequency. Moreover, ultrasonic anemometers suffer less from freezing or mechanical
failures (Anjan et al., 2013).
Both anemometers suffer from mast induced lee effect, meaning the effect of the
mast itself on the wind field. For this reason the measurement setup standards,
regarding for example boom dimensions, should be followed when mounting both
instruments.
21
CHAPTER 2. BACKGROUND
2.4.2
SODAR
SODAR (SOnic Detection And Ranging) is a method for the remote sensing of
wind speed and direction. It is based on the propagation of sound waves through
the atmosphere. Atmospheric turbulence causes temperature fluctuations which
are sensed by SODAR to obtain wind information. The speed of sound is a
function of temperature and humidity which means the acoustic refractive index
of air changes accordingly. Changes in the refractive index cause acoustic energy
backscatter which can be measured using a SODAR. Using timing information
from each scan and the observed Doppler shift of the echoes the wind vector can
be estimated along the measurement path. The measured wind is therefore the
radial component of the true wind vector. The three dimensional wind vector is
then estimated using mathematical algorithms (Bradley, 2010).
A SODAR measurement equipment consists of an emitter which transmits acoustic pulses and a receiver which listens for echoes. In a monostatic setup a single
SODAR antenna performs both sending and receiving functions while a bistatic
setup uses separate antennas. SODARs generally have a single beam pointed
upwards and two or more beams tilted slightly from the vertical and pointed in
orthogonal directions (see Figure 2.7). The vertical beam is used to determine
the vertical structure of the atmosphere. Using geometric calculations the profile of horizontal wind speed and direction can be obtained from the orthogonal
beams. The beams are produced either by using a set of antennas or by a phased
array setup. A phased array uses an array of transducers which are electronically
operated to produce the desired beam shapes. This is achieved by modifying the
interference pattern produced by the transducers.
Doppler SODAR typically measures with a frequency of approximately 1000 Hz
up to 4000 Hz and can be used to obtain vertical wind profiles with a good vertical
and temporal resolution. SODARs typically used in wind power applications have
vertical measurement range of 50-200 meters with 5 meter interval. However, wind
information must be calculated from the obtained signals using mathematical
inversion, which means the interpretation of the SODAR measurements can be
challenging. For instance, temperature and atmospheric humidity must be know a
priori in order to estimate the speed of sound. Additionally, acoustic interference
22
CHAPTER 2. BACKGROUND
Figure 2.7: SODAR measuring principle (FAS).
and clutter (echo from non-moving objects) can cause measurement bias. Falling
raindrops add bias to the wind speed extraction and some SODAR manufacturers
recommend that precipitation periods are filtered out.
Signal-to-noise ratio (SNR) is an important quantity in any remote sensing application. It describes the amount of useful information in the received signal
compared to the ambient noise. In the case of SODAR measurements, examples
of noise sources include nearby wind turbines, wind induced ambient noise and
precipitation. In extremely well mixed boundary layer temperature fluctuations
are small and nearly no sound is reflected back. These situations can be identified
by low SNR (Crescenti, 1998).
The acoustic power received by SODAR as a function of properties of the equipment and atmosphere is described by the SODAR equation:
PR = PT GAe σs
cτ e−2αz
2 z2
(2.15)
where PR is the received power, PT is transmitted power, G is antenna gain
(transmitting efficiency), Ae is antenna (effective) area, σs is target scattering
cross section, c is sound speed in air, τ is pulse duration, α is absorption coefficient
of air and z is height. The equation shows that the measurement range can be
23
CHAPTER 2. BACKGROUND
increased by increasing antenna gain, output power or antenna size. Increasing
pulse duration increases the received power but decreases vertical resolution. The
received power decreases quickly with altitude which limits the vertical reach of
SODARs (Burton et al., 2008).
Usually meteorological sensors are calibrated in a laboratory setting. However,
SODAR calibration is difficult since the instrument measurement volume does
not fit in to a standard wind tunnel. A common way to calibrate a SODAR is
to mount it close (with a minimum separation distance equal mast height) to a
meteorological mast and compare concurrent measurements. Naturally, this only
results in accuracy of equal to or less than the (calibrated) anemometers.
SODAR as a measurement method has several advantages compared to a meteorological mast. A ground based SODAR does not affect the air flow like mast does
and therefore does not interfere with the measured wind. SODAR measurements
do not suffer from under-speeding in freezing conditions or from over-speeding in
turbulent conditions. Both vertical and horizontal wind speeds are available from
SODAR measurements which allows the estimation of flow inclination.
Siting
SODAR measuring performance can be greatly enhanced by proper siting. SODAR
equations assume homogeneous flow inside the measurement volume, and if this
condition is not met the measured winds are skewed depending on the ground
topography. For this reason SODAR should be installed on flat terrain with uniform surface roughness. The location should also be far enough from obstructions
affecting the wind flow such as trees, buildings and wind turbines which affect the
wind flow and cause clutter in the received signal. The device needs to be level,
since even small deviations from horizontal plane will result in error in calculating
the three wind components (Moore and Bailey, 2011).
24
CHAPTER 2. BACKGROUND
2.4.3
Comparing mast and SODAR measurements
Mast and SODAR measurements differ in terms of measuring technique which
makes the comparison between the two challenging. Anemometers measure wind
speed as a scalar quantity whereas SODAR measurements contain wind vector
information. This distinction is important to take into account when comparing
averaged wind speed measurements. Vector-averaged wind can differ from the
scalar-averaged values when there is large variability in wind direction. For example, if wind directions are equally distributed in the sampling window, the wind
vector may average to zero while the scalar wind speed averages to a non-zero
value.
Volume-averaging present in the SODAR measurement introduces a specific issue
to the measurement comparison. Since wind speed profile is generally logarithmic
with respect to height, the volume-averaged measurement centered at height h
is less than the point measurement at the same height. This is due to the shear
in the vertical wind profile. This effect is most noticeable near the surface where
the wind shear is largest and diminishes with altitude.
Moore and Bailey have listed factors originating from the differing physics between
anemometry and SODAR and given the factors an estimated magnitude (Moore
and Bailey, 2006). Additionally, SODAR should never be situated immediately
next to the mast since a nearby mast would interfere with the SODAR measurements. Hence, mast and SODAR do not measure the exact same airflow even
with recommended siting.
2.5
Long-term correlation of measurement data
Short-term wind measurements do not necessarily represent the true wind regime
at the measurement location. For example, wind in the summer months does not
represent the annual wind, nor does the annual wind represent a decade. Usually, the wind resource must be estimated from relatively limited measurements.
This is typically done by correlating the measurements with suitable long-term
reference data. A widely used method is called the Measure-Correlate-Predict
25
CHAPTER 2. BACKGROUND
(MCP) method. Overlapping measurements and reference data are correlated
and the relation is applied to extend the measurements to cover the reference
data period. The method assumes that the calculated correlation is also valid
during the non-overlapping period. The result is an estimated wind time series
for the full temporal extent of the reference data (Romo Perea et al., 2011).
MCP method estimates the wind parameters individually for each input time
series. For this reason the wind speed profile shape only represents the measured
period. From this it follows that the wind profile for example from seasonal
measurements can not be generalized for the full year. The reference data set is
identical for all measurements levels, which implies that the wind speed profile
does not change with time. This has implications particularly when extrapolating
wind speeds to turbine hub height with only seasonal wind profile information.
Using profile shape information from a temporally limited correlation may yield
incorrect extrapolated wind speed values.
26
Chapter 3
Measurements
The meteorological data used in this study contains measurements collected at 17
sites located in both coastal and inland (over 20 km from coast) Finland. 12 of
the sites are classified as inland while the rest are coastal. The exact locations of
the measurement sites are confidential. The measurement campaigns conducted
at the sites varied from single mast or SODAR measurements to multiple mast
and/or SODAR measurements. Total of 265 months of mast data and 116 months
of SODAR data between June 2009 and July 2013 was analyzed (see Table 3.1
and Figure 3.1).
All the sites are located on rather flat terrain. In terms of roughness many of the
sites can be regarded as heterogeneous due to varying forest, agricultural and in
some cases water surfaces.
27
CHAPTER 3. MEASUREMENTS
Table 3.1: Available measurement data. Distance is given for possible concurrent measurements between different measurements at a site.
Site ID
Equipment
Measurement
period
A
A
A
A
B
B
B
C
C
Mast
SODAR1
SODAR2
SODAR3
Mast
SODAR1
SODAR2
Mast
SODAR
D
D
D
E
E
F
F
F
Mast1
Mast2
SODAR
Mast
SODAR
Mast1
Mast2
Mast3
F
SODAR
11/2010-12/2012
11/2011-4/2012
5/2012-8/2012
8/2012-12/2012
7/2012-12/2012
1/2011-10/2012
6/2011-3/2012
4/2011-7/2012
2/2012-3/2012
5/2012-7/2012
11/2009-4/2010
10/2011-9/2012
10/2011-9/2012
9/2009-8/2010
4/2011-7/2011
6/2009-6/2010
4/2011-1/2012
4/2011-12/2011
6/2012-7/2012
6/2012-7/2012
G
G
H
H
H
H
I
I
J
J
K
L
M
N
O
P
Q
Mast
SODAR
Mast1
Mast2
SODAR1
SODAR2
Mast
SODAR
Mast
SODAR
Mast
Mast
Mast
SODAR
SODAR
SODAR
SODAR
10/2010-5/2012
12/2011-5/2012
8/2009-8/2011
11/2010-9/2012
7/2010-9/2010
6/2012-9/2012
12/2011-8/2012
12/2011-8/2012
9/2009-9/2011
1/2011
4/2012-7/2013
4/2011-4/2013
1/2009-1/2010
3/2012-8/2012
11/2011-10/2012
3/2012-10/2012
12/2010-5/2011
Concurrent mast
and SODAR
measurements
x
x
x
x
x
x
x
x
Distance
between
equipment [m]
see below
3800 (S1-M)
5400 (S2-M)
3400 (S3-M)
see below
350 (S1-M)
3850 (S2-M)
see below
200 (S-M)
Distance
from
coast [km]
12
14
12
13
4
4
5
3
3
x
see below
1050 (S-M1)
see below
100 (S-M)
8650 (M2-M3)
see below
12
12
12
47
47
0
3
9
150 (S-M3),
8550 (S-M2)
see below
200 (S-M)
see below
15950 (M1-M2)
10100 (S1-M1)
100 (S2-M2)
see below
50 (S-M)
see below
800 (S-M)
-
9
x
x
x
x
x
x
x
x
x
x
x
x
x
-
28
4
4
18
3
9
3
40
40
1
2
110
5
9
5
160
130
4
CHAPTER 3. MEASUREMENTS
Figure 3.1: Available mast and SODAR measurement periods.
3.1
Mast measurements
Mast heights varied between 50 and 120 meters. The mast measurement levels
were between 40-120 meters and the number of different wind speed measurement
levels was between 1-3. Different mast instrumentations contained different combinations of heated / unheated cup anemometers and ultrasonic anemometers.
Wind direction was measured at least on the top level of wind speed measurements. Air temperature was measured at least on ground (2-3 m height) level.
Other case by case available measurement parameters were air pressure and relative humidity. All the data included 10-minute average, minimum, maximum
and standard deviation values. Mast specific measurement setups are presented
in Table 3.2.
Information on mast instrumentation calibrations was not available.
29
CHAPTER 3. MEASUREMENTS
Table 3.2: Available mast measurement data. Base is the altitude of the base of the
mast.
Site ID
Equipment
A
B
C
D
D
E
F
F
F
G
H
H
I
J
K
L
M
Mast
Mast
Mast
Mast1
Mast2
Mast
Mast1
Mast2
Mast3
Mast
Mast1
Mast2
Mast
Mast
Mast
Mast
Mast
3.2
Measurement
levels (m)
100/70/40
120/82/48
100/70/42
100/70/42
100/70/42
100/70/42
70/57/45
70/57/45
100/70/40
100/70/42
100/70/42
100/70/42
50
100/70/42
82/58/41
120/82/48
100/80/60
Anemometer
model
Thies 1st class
Thies 1st class
NRG #40C
IceFree3/NRG #40C
NRG #40C
IceFree3/NRG #40C
NRG #40C
NRG #40C
NRG #40C
NRG #40C
IceFree3/NRG #40C
NRG #40C
IceFree3
IceFree3/NRG #40C
Thies 1st class
Thies 1st class
IceFree3/NRG #40C
Humidity
x
x
x
x
Height of
base (m)
46
29
23
74
67
68
6
3
44
14
100
21
106
7
261
32
30
SODAR measurements
Available SODAR measurements included horizontal and vertical wind speeds
and direction measurements on five meter intervals either from 20 meters to 150
meters or from 50 meters up to 200 meters. Horizontal wind speed quality index (related to signal-to-noise ratio) was also reported. SODAR data included
10-minute mean and standard deviation of horizontal and vertical wind speeds,
10-minute mean wind direction and data quality index (except for C-SODAR
and F-SODAR which included only 10-minute mean horizontal wind speed and
wind direction). Additionally, measurements on air temperature, humidity and
pressure were conducted on 10-minute interval.
Information on SODAR calibrations was not available. All SODAR measurements were conducted with AQSystem AQ500 Wind Finder which is a monostatic
SODAR (AQSystem).
30
CHAPTER 3. MEASUREMENTS
Table 3.3: Available SODAR measurement data. Base is the altitude of the SODAR
trailer.
Site ID
A
A
A
B
B
C
D
E
F
G
H
H
I
J
N
O
P
Q
Equipment
SODAR1
SODAR2
SODAR3
SODAR1
SODAR2
SODAR
SODAR
SODAR
SODAR
SODAR
SODAR1
SODAR2
SODAR
SODAR
SODAR
SODAR
SODAR
SODAR
Height range (m)
50-200
50-200
50-200
50-200
50-200
50-200
50-200
50-200
50-200
50-200
50-200
50-200
50-200
50-200
20-150
50-200
50-200
50-200
31
Height of base (m)
66
34
53
31
47
24
54
71
44
15
32
19
107
19
40
254
174
12
Chapter 4
Methods
4.1
Data cleaning
Measurements always need to be monitored for faulty recordings since no equipment is flawless and ambient conditions may occasionally distort the measurements. Measurements can be quality controlled either manually or using filtering
algorithms. A simple automatic filtering can for example search for similar consecutive values and flag them. This method is based on the assumption that
small variations (within the resolution of the recordings) from one measurement
to another are expected in a valid time series.
Measurement data was both filtered automatically and inspected visually to remove erroneous values (due to malfunction, icing and other reasons). The automatic filtering was carried out with the condition of ”if the wind speed value is
equal in four consecutive observations, the observations are removed”. However,
visual inspection of the measurement data is considered the most reliable way to
clean the data. This is because error in measurements may appear in many different ways (slowly deteriorating, ”frozen” values, spikes). Therefore conditional
filtering may be difficult to implement.
Data consistency and plausibility (no unexpected spikes) are the main data reliability and quality indicators. Discrepancies in plotted time series from different
heights may indicate that some of the measurement levels are reporting erroneous
32
CHAPTER 4. METHODS
values.
Mast measurements have several known sources of error. Ice formation on mechanical sensors is the most significant source of error in wind measurements in
Finland. Manual cleaning of data aims to identify false recordings especially in
freezing conditions when cup anemometer may collect ice or snow, stop spinning
freely and report a relatively constant close-to-zero value for a period of time.
The identification is made based on wind speed, relative humidity, temperature
and wind speed standard deviation as well as comparison with other available
measurements. High relative humidity, below zero temperatures and low standard deviation are usually present in freezing conditions causing ice formation. Lee
effect from mast was not accounted for in the data post-processing.
AQS500 SODAR provides a quality index representing the measurement signalto-noise ratio (AQSystem, 2013). This ratio can be used to estimate measurement
reliability. SODAR temperature, humidity and pressure data were not analyzed
in this study.
4.2
Data analysis
4.2.1
Data selection
If the mast had several anemometers on one level, the most suitable was chosen
for the analysis. The anemometer type that was used on most levels and located
on the upwind side (in relation to the dominant wind direction) of the mast
was preferred. This allows a comparable wind profile and minimizes the mast lee
effect. Analyzed anemometer data was measured either with NRG 40C Maximum,
NRG IceFree III or Thies 1st class cup anemometers.
On wind profile analyses only complete profiles with concurrent mast measurements on every measurement height were considered. On SODAR wind profile
analyses the required measurement heights were chosen as the closest heights to
the same site mast measurement levels. If the site only had SODAR equipment,
measurements up to 150 m height were required for profile analyses. On some
33
CHAPTER 4. METHODS
sites, observations from all mast measurement heights were not available with
same model instruments. This can add some uncertainty to wind profile and
shear analysis.
4.2.2
Mast and SODAR measurement comparison
When comparing mast and SODAR measurements the datasets must fulfill a
number of conditions. The data to be compared must be concurrent, meaning
both measurements must have been made within the 10-minute measurement
interval. For wind profile analyses only complete profiles from both measurement
sets were considered. For mast-SODAR comparisons at a certain height the
closest available height from SODAR measurements was considered in relation to
the mast measurements.
In some cases the distance between the mast and the SODAR was over 500 meters
and assumption of homogeneous wind conditions becomes questionable. Despite
the rather large distance, these sites were included in the study because deploying
a SODAR next to a meteorological mast is not always possible or even desirable.
4.3
4.3.1
Terminology
Availability
Data availability is the number of measurements over the maximum possible
number of measurements in a given period taking the measurement time step
into account. Data recovery should be at least 90 % during the measurement
period so that all temporal variability can be included and that average values
are not conditionally biased.
4.3.2
Turbulence intensity
Turbulence intensity (TI) was calculated according to Equation 2.3. Applied wind
speed and wind speed standard deviation were from the highest mast measure34
CHAPTER 4. METHODS
ment level and for SODAR the level closest to this of mast.
4.3.3
Height levels for the calculation of alpha
Power law exponent values (referred to as alpha, see Equation 2.12) for masts
are calculated from the upper two anemometer measurements. When a complementary SODAR is present, the corresponding SODAR alpha values are calculated using the nearest corresponding measurement heights. For sites with only
SODAR measurements, alpha was calculated using 70 m and 100 m heights (most
common top and middle anemometer heights in available mast data).
4.3.4
Season categories
In order to study seasonal effects, three season categories were defined. Winter
covers months of December, January and February while summer includes June,
July and August. Spring/fall entails the rest (March, April, May, September,
October and November).
4.3.5
Regression analysis parameters
Slope and offset are parameters obtained from applying a least-squares linear fit
to the mast and SODAR data. If the linear fit is in the form of:
y = ax + b
(4.1)
then the linear fit coefficient a defines the slope parameter while the coefficient b
defines the offset parameter. The least-squares estimates of the slope and offset
parameters are then
a=
Pn
(x −x̄)(yi −ȳ)
i=1 i
P
n
2
i=1
b = ȳ − ax̄
35
(xi −x̄)
(4.2)
CHAPTER 4. METHODS
The goodness of the fit is described by the coefficient of determination (R2 ). This
parameter is obtained from the residuals of the least-squares regression:
R2 = 1 − SSres
tot
P
Stot =
(yi − ȳ)2
P
Sres =
(yi − fi )2
(4.3)
Here yi are the observations and fi are the corresponding values estimated using
the regression (Draper et al., 1966).
36
Chapter 5
Results
5.1
5.1.1
Data processing
Data cleaning
Quality control procedures should be applied to measurement data before use.
The most noticeable decrease in raw data quality in this study was likely due to
ice build-up on anemometers. An example of the effect of freezing conditions on
anemometer measurements is presented in Figure 5.1.
Figure 5.1: An example of wind speed time series from mast M. Discrepancies between
the measurements imply ice build-up on the cup anemometers.
37
CHAPTER 5. RESULTS
Conditional filtering was attempted in this study, but was eventually replaced
with manual inspection. The attempted filtering was based on detecting similar
consecutive values and was deemed insufficient. Due to the diversity of the freezing conditions’ manifestation in anemometer time series (example in Figure 5.1),
the errors can be very difficult to automatically identify and a much more advanced filtering method would have been necessary. Data was cleaned according
to the manual method presented in Section 4.1.
5.1.2
Measurement data quality
SODAR data was filtered using the quality index value of 40. In Figure 5.2 an
example of the occurrence of low quality measurements is presented as all of the
cup anemometers experience ice build-up. All analyses regarding SODAR data
quality can be found in Section 7.1. The analyses were performed seasonally for
sites with concurrent mast measurements.
The highlighted low quality measurements in Figure 5.2 exhibit greater scatter
especially near the higher end of the wind speed spectrum. This was also evident in the rest of the analyses. In summer-time the unreliable measurements
tend to occur in low wind speeds. SODAR is expected to perform poorly in well
mixed boundary layer due to weak scattering by the minimal temperature fluctuations. In these cases relatively low wind speeds are common. In winter-time
the unreliable measurements mostly occur in high wind speeds and exhibit the
greatest scatter. In spring and autumn the scatter is smaller and thus unlikely
to cause significant bias. Low quality in high wind speeds is likely partly caused
by wind-induced ambient noise.
Concurrent mast and SODAR wind speed measurements from available sites
were compared using regression analysis. The resulting key parameters for the
mast-SODAR measurement pairs are presented in Table 5.1 next to the distance
between the measurement equipment and height difference of their locations. The
presented parameters are slope and offset of the linear regression as well as the
coefficient of determination.
Figure 5.3 depicts the effect of data quality control on mast-SODAR comparison.
38
A-SODAR1 wind speed [m/s]
CHAPTER 5. RESULTS
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
A
Winter
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
10
15
20
25
10
15
20
25
Summer
5
A-Mast wind speed [m/s]
Figure 5.2: Regression analysis of mast and SODAR wind speeds with low quality (quality index < 40) SODAR measurements highlighted in red for site A. The empty plot
signifies that no concurrent measurements in summer are available.
Quality control significantly decreased scatter and brought the slope value to a
more reasonable level at nearly all sites. The offset value, while usually closer to
zero than before quality control, can remain relatively high. The remaining bias
can be result of improper SODAR siting, calibration errors or non-comparable
measuring locations, among other reasons. Overall, quality control seems to improve the slope parameter and should be performed before any further analyses.
However, it should be noted that 1 to 1 correspondence when comparing mast
and SODAR may not be realistic due to the reasons stated above. All regression
analyses regarding the effect of data quality control can be found in Section 7.2.
Table 5.1 shows that neither height of equipment base nor distance between the
instruments correlate particularly well with coefficient of determination. This
is likely caused by the lack of directional filtering on mast measurements and
39
CHAPTER 5. RESULTS
Table 5.1: Slope, offset and R2 of the regression analyses of mast-SODAR pairs.
SODAR measurements are the dependent variable.
Measurement pair
Slope
Offset
[m/s]
R2
A-Mast - A-SODAR1
A-Mast - A-SODAR2
A-Mast - A-SODAR3
B-Mast - B-SODAR1
C-Mast - C-SODAR
D-Mast2 - D-SODAR
F-Mast3 - F-SODAR
G-Mast - G-SODAR
H-Mast1 - H-SODAR1
H-Mast2 - H-SODAR2
I-Mast - I-SODAR
J-Mast - J-SODAR
1.04
0.97
1.02
1.00
1.02
0.94
0.92
0.90
0.89
0.98
0.90
0.99
0.55
0.46
0.34
0.34
0.50
-0.27
0.63
0.71
0.02
0.10
0.21
0.40
0.88
0.81
0.87
0.93
0.92
0.86
0.77
0.94
0.77
0.96
0.79
0.91
Distance
between
equipment [m]
3800
5400
3400
350
200
1050
150
200
10100
100
50
800
Difference in
equipment elevation of
base (mast-SODAR) [m]
-20
12
-7
-2
-1
13
0
-1
68
2
-1
-12
varying topography (orography and roughness) of the sites and their immediate
surroundings.
C (100 m height)
C-SODAR wind speed [m/s]
20
15
10
5
00
5
15
10
C-Mast wind speed [m/s]
y = 0.83x + 1.60
y = 1.02x + 0.50
Unfiltered
Filtered
20
Figure 5.3: The effect of quality control on mast-SODAR comparison using linear regression analysis. Blue values are raw data and red values quality controlled data.
40
CHAPTER 5. RESULTS
5.1.3
Data availability
SODAR measurements include a quality index which can be used to estimate
measurement quality. In Figure 5.4 the effect of quality index thresholding on
SODAR data availability is presented. As can be seen from the figure, the chosen
threshold value for quality index (40) seems to be of suitable magnitude without
great loss of availability. Availability starts to decrease rapidly after this value.
Consequently, the SODAR data used in this study have been filtered using quality
index threshold value of 40.
100
90
Availability [%]
80
70
60
50
40
300
A-SODAR1
B-SODAR2
A-SODAR3
A-SODAR2
H-SODAR2
B-SODAR1
G-SODAR
I-SODAR
H-SODAR1
D-SODAR
20
40
60
Quality index threshold
80
100
Figure 5.4: The effect of quality index thresholding on SODAR data availability. X-axis
denotes the minimum accepted quality index value.
The higher the availability, the more the measurement data can be trusted to
represent the true conditions including temporal variability. If data unavailability
is not random but caused by particular atmospheric conditions the resulting data
will be biased. The resulting data availabilities after quality control are presented
in Table 5.2.
41
CHAPTER 5. RESULTS
Table 5.2: Data availabilities of the main measurement parameters after quality control.
Wind speed availabilities are presented for all mast measurement heights and for the
closest corresponding SODAR measurement heights. For SODARs without a mast pair
only the heights used in alpha calculations are presented.
Equipment
A-Mast
A-SODAR1
A-SODAR2
A-SODAR3
B-Mast
B-SODAR1
B-SODAR2
C-Mast
C-SODAR
D-Mast1
D-Mast2
D-SODAR
E-Mast
E-SODAR
F-Mast1
F-Mast2
F-Mast3
F-SODAR
G-Mast
G-SODAR
H-Mast1
H-Mast2
H-SODAR1
H-SODAR2
I-Mast
I-SODAR
J-Mast
J-SODAR
K-Mast
L-Mast
M-Mast
N-SODAR
O-SODAR
P-SODAR
Q-SODAR
Speed
(top)
97
97
97
86
91
95
96
97
98
97
97
97
93
98
93
97
99
89
95
97
97
89
99
82
91
96
95
96
83
92
98
99
97
98
97
Speed
(middle)
97
99
98
88
91
98
99
97
99
98
97
98
92
99
93
96
99
90
93
98
95
89
100
83
98
94
98
83
95
99
99
98
98
98
Speed
(bottom)
97
100
99
88
92
99
99
98
99
82
97
98
77
99
93
94
98
90
93
98
85
88
100
83
99
91
99
83
95
91
-
Direction
(top)
99
97
97
86
98
95
96
99
98
97
96
97
98
98
94
98
99
89
100
97
97
98
99
82
73
96
97
96
99
100
96
99
97
98
97
Temperature
(ground)
99
90
100
100
100
100
90
100
82
99
100
99
100
100
99
100
100
-
Humidity
(ground)
99
17
99
100
-
Availabilities are generally good (>90 %) except in a few cases. The majority
of rejected data was removed due to suspected anemometer freezing. For this
reason availabilities during the summer months are better than those in winter.
42
CHAPTER 5. RESULTS
Low availability in the winter months can cause bias in annual mean wind speed,
since wind speed is generally higher during winter. The application of the MCP
method can help to mitigate this effect.
If anemometer freezing is left unidentified, the wind speed average from anemometers is generally lower than the true value. As the quality controlled availabilities remain relatively high, the data filtering could have been even more aggressive.
This would help to mitigate the conditional bias due to faulty anemometer measurements.
Mast lee effect was not taken into consideration in the data post-processing due
to lack of information on boom orientations. This may lower the recorded wind
speed values from anemometers mounted on horizontal booms. Wind profiles are
compiled from anemometer measurements located at the same side of the mast.
Lee effect should therefore occur simultaneously in all anemometers, provided
that wind direction is equal at all measurement heights.
5.2
Atmospheric stability conditions
The most widely applied formulas for wind profile estimation (e.g. power law)
assume a neutrally stratified boundary layer. Therefore it is necessary to study
how often this assumption holds.
5.2.1
Stability estimation methods
Comparison of the sigma theta and the delta T methods for estimating the P-G
stability categories according to the Table 2.1 was performed. There were three
sites with sufficient measurements for both methods. In Figure 5.5 the comparison
for site K is presented. All comparisons can be found in Section 7.3.
The categories produced by the sigma theta method seem to be normally distributed with mode at category E. The delta T method on the other hand has
a secondary peak at category A. The reason for this could be that the delta
T method is more sensitive to thermal turbulence. This supports the use of the
43
CHAPTER 5. RESULTS
60
Occurrence [%]
50
K-Mast
Delta T
Sigma theta
40
30
20
10
0 A
B
D
E
C
Stability category
F
G
Figure 5.5: Comparison of the sigma theta and the delta T methods in determining the
P-G categories for site K.
sigma theta method as the primary method for stability estimation. Additionally,
the data required for the sigma theta method are usually available from meteorological mast measurements contrary to more complex methods. For these reasons
this study uses the sigma theta method exclusively for stability estimations.
5.2.2
Stability statistics based on measurements
Occurrence of different atmospheric stability conditions according to the sigma
theta method presented in Table 2.1 were determined (see Table 5.3). Stability
categories could only be estimated for the mast measurements, since SODARs
did not provide the necessary information. For C-SODAR and F-SODAR measurements, the stability category could not be determined due to the absence of
wind direction standard deviation measurements. It is important to note that the
presented stability distribution represents only the statistics of the measurement
periods. As some sites do not have an entire year of measurements the distributions may not be representative of the annual conditions. For the measurement
periods for each site please see Table 3.1.
The presented stability distributions are typical for Finnish conditions where
neutral and slightly stable stability categories are most common. It should be
44
CHAPTER 5. RESULTS
Table 5.3: Occurrence of different Pasquill-Gifford stability categories calculated with
the sigma theta method. Values are percentages of the total number of measurements.
Location
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Inland
Inland
Inland
Equipment
A-Mast
B-Mast
C-Mast
D-Mast1
D-Mast2
F-Mast1
F-Mast2
F-Mast3
G-Mast
H-Mast1
H-Mast2
J-Mast
L-Mast
M-Mast
E-Mast
I-Mast
K-Mast
A
1
1
2
0
0
2
1
4
1
1
0
1
2
1
1
4
2
B
1
1
2
0
0
5
1
1
1
1
1
1
1
1
1
5
2
C
5
3
5
2
2
28
8
4
4
5
3
4
6
4
4
18
4
D
27
14
36
20
20
41
48
20
35
25
28
27
28
29
18
42
15
E
40
48
31
46
38
16
27
46
37
39
40
41
39
37
42
17
47
F
13
20
6
10
9
3
5
7
7
8
7
8
14
8
11
5
18
G
12
13
17
22
31
5
10
17
17
21
20
19
9
20
23
9
12
noted that most sites exhibit relatively large proportion of stability category G.
This is likely caused by the relatively large bin range of the category.
The yearly cycle of atmospheric stability for the analyzed sites is presented in
Table 5.4. As is the case with Table 5.3, the stability categories could be estimated only for mast measurements. The mean category is obtained by taking the
monthly mean of the wind direction standard deviation and converting the result
to P-G stability category.
Based on site-specific results, the most common monthly stability category is E
(67 %) followed by D (33 %) and G (<1 %). Winter months are generally more
stable compared to summer months, as is expected. The results presented in
Tables 5.3 and 5.4 do not show any clear divide between coastal and inland sites
in terms of stability categories.
45
CHAPTER 5. RESULTS
Table 5.4: Mean monthly Pasquill-Gifford stability categories calculated with the sigma
theta method. Dashes indicate months with unavailable measurements.
Location
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Inland
Inland
Inland
5.3
Equipment
A-Mast
B-Mast
C-Mast
D-Mast
D-Mast2
F-Mast1
F-Mast2
F-Mast3
G-Mast
H-Mast1
H-Mast2
J-Mast
L-Mast
M-Mast
E-Mast
I-Mast
K-Mast
Jan
E
E
E
E
D
E
E
E
E
E
E
E
E
D
E
Feb
E
E
E
E
D
E
E
E
E
E
E
F
D
E
Mar
E
D
E
E
D
E
E
E
E
E
G
E
E
D
E
Apr
E
D
E
E
D
D
E
E
E
E
E
D
E
E
D
E
May
D
D
E
D
D
E
D
D
E
E
D
E
D
D
D
Jun
D
D
E
D
D
E
E
D
E
D
D
D
D
D
D
Jul
D
E
E
E
D
D
D
E
D
E
E
D
E
E
D
E
Aug
E
E
E
E
D
D
D
E
E
E
E
D
E
E
D
E
Sep
E
E
D
E
D
D
D
E
E
E
E
E
E
E
E
Oct
E
E
D
E
D
D
D
E
E
E
E
E
E
E
E
Nov
E
E
E
E
E
D
D
D
E
E
E
E
E
E
E
E
Dec
E
E
D
E
E
D
D
E
E
E
E
E
E
E
E
D
E
Evaluation of SODAR reliability
SODAR measurements were compared to mast measurements in various atmospheric conditions in order to study the factors contributing to SODAR reliability.
Quality controlled mast measurements are taken to represent the true wind conditions and are used as a reference for SODAR measurements.
In order to estimate the reliability of SODAR measurements, a variety of parameters were analyzed relative to mast-SODAR wind speed difference. In Figure
5.6 the mast-SODAR wind speed difference is plotted against SODAR-measured
wind speed. Analyses for all sites can be found in Section 7.4. It can be seen
that SODAR overestimation of wind speed increases with SODAR-measured wind
speed. This can be caused by loud ambient noise in high wind conditions. Measurements in high wind conditions should particularly be observed for low signalto-noise ratio and discarded if appropriate. The resulting gaps can be filled using
the MCP-method. This way the mean wind speed is not affected by uncertain
measurements. Following from the overestimation, the SODAR-calculated alpha
is increasingly overestimated in high wind speeds. Likewise, wind profile estima46
CHAPTER 5. RESULTS
Wind speed difference (A-Mast - A-SODAR1) [m/s]
tions are increasingly biased with height.
A (100 m height, N = 20527)
Median
5
0
5
0
5
15
10
A-SODAR1 wind speed [m/s]
20
Figure 5.6: Mast-SODAR wind speed difference as a function of SODAR wind speed
for site A, SODAR 1.
Figure 5.7 presents the wind speed difference as a function of turbulence intensity
calculated based on mast measurements. SODAR is known to measure unreliably
in low turbulence which is usually characteristic for low wind speed. The dip in
median wind speed difference, observed at many sites at low turbulence intensity
values, is in support of that.
Wind speed difference as a function of humidity is shown in Figure 5.8. In two
out of the three comparisons the measurements showed increasing scatter in wind
speed difference towards greater humidity values. However, the median difference remains rather constant. The uncertainty in SODAR wind speeds at high
humidity conditions can be caused by rain as observed in earlier studies.
Mast-SODAR wind speed difference was studied in relation to alpha and stability
category but no noticeable correlation was found.
47
Wind speed difference (D-Mast2 - D-SODAR) [m/s]
CHAPTER 5. RESULTS
D (100 m height, N = 44913)
Median
5
0
5
0
10
20
30
40
50
D-Mast2 turbulence intensity [%]
60
70
Wind speed difference (A-Mast - A-SODAR1) [m/s]
Figure 5.7: Mast-SODAR wind speed difference as a function of mast calculated turbulence intensity for site D.
A (100 m height, N = 20527)
Median
5
0
5
20
30
40
50
60
70
80
A-Mast relative humidity [%]
90
100
Figure 5.8: Mast-SODAR wind speed difference as a function of mast measured relative
humidity for site A, SODAR 1.
48
CHAPTER 5. RESULTS
5.4
Wind speed vertical profile
Understanding the behavior of the vertical wind profile in different atmospheric
conditions enables more accurate hub height wind speed estimations. Wind profile
depends on atmospheric stability which in turn correlates with season.
5.4.1
Stability dependency
In Figure 5.9 an example of the dependency of mast alpha values on stability
category is presented. For figures on the rest of the mast measurements the
reader is directed to Section 7.7.
Alpha
C-Mast
N
25
Sample size (thousands)
Alpha (Mast, 70 m - 100 m)
1.5
1.0
20
0.5
15
0.0
10
0.5
5
1.0A
B
C
D
E
Stability category
F
G0
Figure 5.9: Mast alpha as a function of P-G stability category for site C. The red line
denotes the measurement sample size.
A more stable boundary layer exhibits higher alpha values as expected. The drop
in alpha values at category G is likely caused by very low wind speed values
typical in extremely stable conditions. In these conditions the entire mast can be
within the nearly windless stable layer and therefore wind speed measurements
at the alpha levels are close to zero. Estimating alpha values from near-zero
measurements is of questionable usefulness.
49
CHAPTER 5. RESULTS
5.4.2
Seasonal variability
The wind speed profile shape was studied site-specifically dividing all available
measurements to three season categories (for definition see Section 4.3.4). Figure
5.10 shows an example of seasonal wind profile shapes from various co-locating
instruments. The profile gradient changes between the seasons and is steepest
during winter and lowest during summer. It should be noted that the profiles are
not necessarily concurrent between the instruments.
F
Other
Summer
120
120
100
100
100
80
60
F-Mast1
F-Mast2
F-Mast3
F-SODAR
40
203
4
5
6
7
Wind speed [m/s]
8
Height [m]
120
Height [m]
Height [m]
Winter
80
60
F-Mast1
F-Mast2
F-Mast3
F-SODAR
40
9
203
4
5
6
7
Wind speed [m/s]
8
80
60
F-Mast1
F-Mast2
F-Mast3
F-SODAR
40
9
203
4
5
6
7
Wind speed [m/s]
8
9
Figure 5.10: Average seasonal profiles for the masts and SODAR at site F. Each profile
represents the average of all complete profiles available in the given season. The difference in wind profile shape between the seasons can be observed from the mast profiles.
Figure 5.10 shows that using seasonal wind profile shape outside the season yields
incorrect extrapolated hub height wind speeds. If measurements do not cover full
years (N * 12 months), the retrieved alpha can vary a lot from the annual average.
Seasonal variability of wind speed and alpha were studied by calculating monthly
wind speed and alpha relative to the mean for masts with a minimum of 12
months of measurements (see Tables 5.5 and 5.6). Top level mast data was used
for monthly wind speed statistics. Alpha levels were chosen using the definitions
in Section 4.3.3.
Both monthly wind speed and alpha ratios exhibit the expected cyclical pattern
50
CHAPTER 5. RESULTS
Table 5.5: Calculated monthly wind speeds relative to mean annual wind speed for sites
with a minimum of 12 months of measurements. The mean annual wind speed can be
found in the last column. The last row shows the site-averaged monthly ratio.
Location Equipment
Jan
Feb
Mar
Apr
May Jun
Jul
Aug
Sep
Oct
Nov
Dec
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Inland
Inland
Inland
1.07
1.07
1.03
0.99
1.08
1.14
1.15
1.06
1.13
1.07
1.11
1.17
0.75
1.02
1.08
1.03
1.13
0.98
1.04
1.00
1.06
1.06
0.96
1.08
0.96
1.11
1.01
1.15
1.10
1.04
1.18
1.18
1.08
1.05
1.15
1.10
1.17
1.16
1.10
0.98
1.16
1.07
1.13
0.98
1.12
0.93
0.95
0.99
0.94
0.92
0.88
0.94
0.94
0.97
0.96
0.92
0.95
1.06
0.95
0.95
0.98
1.02
0.98
0.87
0.95
0.97
1.03
0.98
1.06
1.00
0.97
0.90
1.08
1.04
1.00
0.85
0.89
0.91
0.93
0.80
0.87
0.82
0.94
0.82
0.89
0.83
1.04
0.95
0.88
0.88
0.84
0.90
0.82
0.99
0.86
0.87
0.83
0.95
0.81
0.94
0.88
0.95
0.83
0.81
0.87
1.08
1.11
1.08
1.21
1.03
1.06
1.03
1.08
1.05
1.11
1.05
1.17
1.09
1.05
1.08
1.02
1.20
1.08
1.01
1.10
1.08
1.14
1.13
1.05
0.95
1.11
0.93
1.03
0.96
1.07
1.12
1.12
1.02
1.14
1.03
1.02
1.15
1.02
1.17
1.08
1.23
0.99
1.23
1.18
1.10
1.10
1.32
1.28
1.10
1.11
1.13
1.12
1.10
1.16
1.08
1.27
1.12
1.05
1.18
1.14
A-Mast
C-Mast
D-Mast2
F-Mast1
G-Mast
H-Mast1
H-Mast2
J-Mast
L-Mast
M-Mast
B-SODAR1
E-Mast
K-Mast
O-SODAR
Mean
0.82
0.83
0.89
0.97
0.80
0.92
0.88
1.00
0.86
0.86
0.88
0.99
0.86
0.85
0.89
Mean
[m/s]
6.17
5.90
6.60
5.10
6.40
6.70
6.50
6.30
6.58
6.30
7.22
6.10
6.32
6.70
6.35
Table 5.6: Calculated monthly power law alphas relative to mean annual alpha for sites
with a complete 12-month measuring period. The mean annual alpha can be found in
the last column. The last row shows the site-averaged monthly ratio.
Location
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Coastal
Inland
Inland
Inland
Equipment
A-Mast
C-Mast
D-Mast2
F-Mast1
G-Mast
H-Mast1
H-Mast2
J-Mast
L-Mast
M-Mast
B-SODAR1
E-Mast
K-Mast
O-SODAR
Mean
Jan
1.05
1.30
1.22
0.97
1.24
1.04
0.99
1.26
0.94
1.06
1.07
1.04
1.75
1.19
1.12
Feb
0.86
0.75
0.89
0.77
1.11
1.10
0.74
0.99
0.99
1.18
1.21
0.99
1.09
1.01
1.00
Mar
1.03
0.92
1.08
0.94
1.03
0.80
1.11
1.20
0.98
0.99
1.21
1.10
1.20
1.12
1.05
Apr
0.88
0.74
0.96
0.76
1.06
0.47
0.85
0.93
0.90
0.84
1.15
0.66
1.01
0.93
0.88
May
0.81
0.81
0.83
0.88
0.96
0.20
0.71
0.69
0.79
0.76
0.90
0.36
1.11
0.96
0.77
Jun
0.67
0.65
0.80
1.00
0.99
0.60
0.62
0.60
0.74
0.53
0.75
0.52
0.93
0.98
0.73
Jul
0.79
0.85
0.84
0.65
1.22
0.70
0.62
0.61
0.74
0.86
0.78
0.80
1.00
1.04
0.80
Aug
0.69
0.89
0.86
0.93
0.90
1.14
0.79
0.59
0.79
0.96
0.82
0.87
0.95
1.04
0.86
Sep
0.98
0.86
1.08
0.84
0.98
1.43
1.00
0.81
0.98
1.02
0.99
1.14
1.14
1.10
1.02
Oct
0.98
0.85
1.11
0.98
1.05
1.79
1.22
0.93
0.97
1.20
1.06
1.02
1.38
1.41
1.11
Nov
1.08
0.83
1.14
0.96
1.20
1.14
1.26
1.20
1.10
1.04
1.21
0.96
1.24
1.11
1.12
Dec
1.08
1.22
1.19
1.08
1.16
1.18
1.37
1.15
0.88
1.10
1.17
1.31
1.02
1.13
1.14
Mean
0.33
0.39
0.33
0.48
0.45
0.36
0.28
0.32
0.39
0.55
0.38
0.20
0.33
0.40
0.37
reasonably well. Monthly average wind speed ratio varied from 87 % in August
to 114 % in December. The range of average monthly alpha ratio was from 73 %
in June to 114 % in December.
The seasonal variability of alpha poses a problem when estimating the hub height
wind speeds based on a short measurement campaign. A solution is proposed
which entails using the seasonal wind speed and alpha from measurements and
correcting them with the presented monthly statistics (Tables 5.5 and 5.6). The
51
CHAPTER 5. RESULTS
correction is done in the following manner:
uannual = useasonal /C
αannual = αseasonal /D
(5.1)
where the seasonal values are the mean of the short-term measurements and C
and D are the mean correction factors (see Tables 5.5 and 5.6) corresponding to
the months in question.
After this, wind speed at the desired hub height can be calculated using the
power law wind profile. The method as a whole is described in Figure 5.11.
Figure 5.12 shows the effect of alpha and wind speed correction on hub height
wind speed extrapolation. As the method is based on limited wind statistics,
the result should be used only as indicative and to assess the uncertainty due to
vertical extrapolation.
52
CHAPTER 5. RESULTS
Seasonal wind
measurements
Seasonal alpha
Seasonal wind speed
Statistical correction
of seasonal values
Annual wind speed
Annual alpha
Hub height wind speed
extrapolation
Annual hub height
wind speed
Figure 5.11: The proposed method to estimate annual hub height wind speed using
seasonal measurements.
53
CHAPTER 5. RESULTS
Seasonal
Height
Annual
Wind speed
Wind speed
Figure 5.12: Illustration of the annual hub height wind speed extrapolation method.
Firstly, seasonal wind speed and alpha are calculated from measurements. After this
the corresponding annual values are estimated using the statistics. The resulting annual
wind speed and alpha are used to extrapolate the hub height wind speed. Orange dots
indicate the extrapolated hub height wind speeds before and after the correction.
54
Chapter 6
Conclusions and discussion
This study analyzed 265 months of mast measurement data and 116 months of
SODAR measurement data from 17 sites in Finland. The measurement campaigns varied greatly with differing amount and type of equipment, temporal
data coverage and measurement concurrency. Sites with concurrent mast and
SODAR measurements had greatly varying distances between the measurement
locations. This limited the applicability of certain analytic methods. However,
these variabilities are far from uncommon and therefore reflect the true state of
wind resource assessment in wind power applications.
After the data quality control procedures the data recovery remained high and
often exceeded 90 %. This encourages even more rigorous post-processing measures without risking data availability. While automatic measurement filtering
was attempted, manual data inspection remains the most useful and effective
way for measurement quality control. This study indicates that an effective automatic data filtering, while possible, can be challenging to implement and needs
far more dedicated research beyond the scope of this study.
The sites were classified to different stability categories using the sigma theta
method. Neutral and slightly stable stability categories were the most common
by a large margin. The sigma theta method seems adequate and is recommended
for determining the P-G stability categories due to the simplicity of the required
input data. No noticeable difference between coastal and inland sites in terms of
55
CHAPTER 6. CONCLUSIONS AND DISCUSSION
stability variability was found.
Stable conditions exhibit greater vertical wind shear and thus larger alpha values
than neutral and unstable conditions. This has implications when determining
time-averaged alpha. SODAR has a tendency to lose measurements in well mixed
boundary layer where low alpha values are common. In conclusion, SODAR can
overestimate the annual mean alpha.
The absolute accuracy of SODAR measurements could not be estimated due to
great distances between most of the mast-SODAR pairs. In addition, incomplete
information on anemometer calibration and mast boom orientations as well as
SODAR calibration and siting inhibited accuracy comparisons. Nevertheless, the
dependency of mast-SODAR wind speed bias on different prevailing atmospheric
conditions was assessed in multiple comparisons.
SODAR overestimation in high wind speeds was observed at most sites which
leads to e.g. unrepresentative wind profiles. Mast-SODAR wind speed difference
was found to generally increase with diminishing turbulence intensity. Based
on the few comparisons of wind speed difference against humidity the scatter in
wind speed difference increases with humidity. However, no significant trend was
observed. The results point to the fact that humidity plays a negligible role in
SODAR accuracy except perhaps very near to 100 %.
The variability of wind speed profiles was analyzed seasonally. The profiles vary
significantly with season in terms of wind speed and shear. Additionally, monthly
wind speed and alpha statistics were studied and both showed a clear seasonal
cycle. Therefore hub height wind speed extrapolations should not be performed
based on purely seasonal data unless seasonal features are taken into account.
One method how to perform this is presented in this study. The method can also
be used to estimate wind speed extrapolation uncertainty.
Using the obtained statistics, annual wind speed and alpha values can be generalized from the seasonal ones. Applying the generalized values the hub height wind
speed can be estimated in a more reliable manner. Since the annual corrections
are based on limited sample size the method should be applied mainly as an indicator of extrapolation accuracy. Uncertainty analysis plays a large role in wind
power project risk management. The proposed method presents an applicable
56
CHAPTER 6. CONCLUSIONS AND DISCUSSION
tool for reaching a quantitatively justified estimation of the uncertainty related
to a short measurement campaign in Finnish conditions.
This study shows that short SODAR measurement periods (less than 3 months)
can be unreliable for various reasons when the goal is to determine average annual conditions. Measurement periods which experience rainfall (as indicated
by near 100 % relative humidity) are susceptible to poor measurement quality.
In low turbulence periods SODAR measurement uncertainty is also increased.
Measurement periods with these conditions prevailing should be avoided.
Timing the measurement campaign to winter may lead to increased uncertainty
in high wind speeds as winters are windier and SODAR wind speed uncertainty
was found to increase with wind speed. Then again summer measurement periods
may experience higher uncertainties in low wind speeds as well mixed boundary
layer with weak winds are relatively common. Since the wind speed range of
energy production in utility scale wind turbines is approximately 3-30 m/s and
the nominal power with e.g. a 3 MW turbine is achieved typically at wind speeds
above approximately 10 m/s, the accurate estimation of the high end of the wind
speed spectrum is more important in wind power applications. Consequently,
short SODAR measurement campaigns should be conducted outside the winter
season.
Excluding low wind speeds could be justified in wind shear examinations in wind
power motivated studies regarding wind profile. This stems from the fact that
most wind turbines do not produce power with wind speeds less than 3 m/s.
As stated in International Electrotechinal Commission, 2005, calibration of cup
anemometers in wind power applications is required only at wind speeds between
4 m/s and 16 m/s.
While data cleaning yielded satisfactory results, more thorough measurement
data filtering and stricter criteria could be applied without the fear of low data
availability. Efforts towards automatic data filtering would be beneficial since
data inspection is among the most time consuming tasks in wind resource assessment, as was the case in this study. For more reliable mast-SODAR measurement
comparisons the mast data processing should account for the mast lee effect.
Before using SODAR as a standalone instrument a verification measurement
57
CHAPTER 6. CONCLUSIONS AND DISCUSSION
period against a meteorological mast is recommended. The verification period
should contain as much variability in meteorological conditions as possible. This
ensures a comprehensive assessment of SODAR performance. Simply increasing
the measurement data sample size may not be sufficient if the prevailing meteorological conditions are uncharacteristic.
Temperature measurements from at least two height levels would allow the use
of more complex stability estimation methods. For example, the use of bulk
Richardson number as the stability indicator would be significantly more realistic
than the sigma theta or delta T methods. Additionally, with comprehensive
measurement data, analysis on the performance of different stability estimation
methods could be carried out.
58
Chapter 7
Appendices
59
CHAPTER 7. APPENDICES
A-SODAR2 wind speed [m/s]
A-SODAR1 wind speed [m/s]
7.1
Mast-SODAR regression analysis, part 1
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
A
Winter
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
10
15
20
25
10
15
20
25
15
20
25
Summer
5
A-Mast wind speed [m/s]
A
Winter
5
10
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
A-Mast wind speed [m/s]
20
25
Figure 7.1: Regression analysis of mast and SODAR wind speeds with low quality (quality index < 40) measurements highlighted in red, site A.
60
B-SODAR1 wind speed [m/s]
A-SODAR3 wind speed [m/s]
CHAPTER 7. APPENDICES
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
A
Winter
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
10
15
20
25
15
20
25
A-Mast wind speed [m/s]
B
Winter
5
10
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
B-Mast wind speed [m/s]
20
25
Figure 7.2: Regression analysis of mast and SODAR wind speeds with low quality (quality index < 40) measurements highlighted in red, sites A and B.
61
D-SODAR wind speed [m/s]
C-SODAR wind speed [m/s]
CHAPTER 7. APPENDICES
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
C
Winter
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
C-Mast wind speed [m/s]
20
25
D
Winter
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
D-Mast2 wind speed [m/s]
20
25
Figure 7.3: Regression analysis of mast and SODAR wind speeds with low quality (quality index < 40) measurements highlighted in red, sites C and D.
62
G-SODAR wind speed [m/s]
F-SODAR wind speed [m/s]
CHAPTER 7. APPENDICES
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
F
Winter
5
10
15
20
25
5
10
15
20
25
Other
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
F-Mast3 wind speed [m/s]
20
25
G
Winter
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
10
15
20
25
10
15
20
25
Summer
5
G-Mast wind speed [m/s]
Figure 7.4: Regression analysis of mast and SODAR wind speeds with low quality (quality index < 40) measurements highlighted in red, sites F and G.
63
H-SODAR2 wind speed [m/s]
H-SODAR1 wind speed [m/s]
CHAPTER 7. APPENDICES
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
H
Winter
5
15
10
20
25
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
10
15
20
25
15
20
25
H-Mast1 wind speed [m/s]
H
Winter
5
10
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
H-Mast2 wind speed [m/s]
20
25
Figure 7.5: Regression analysis of mast and SODAR wind speeds with low quality (quality index < 40) measurements highlighted in red, site H.
64
J-SODAR wind speed [m/s]
I-SODAR wind speed [m/s]
CHAPTER 7. APPENDICES
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
25
20
15
10
5
00
I
Winter
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Other
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
20
25
Summer
SODAR DQI ≥ 40
SODAR DQI < 40
5
15
10
I-Mast wind speed [m/s]
20
25
J
Winter
SODAR DQI ≥ 40
SODAR DQI < 40
5
10
15
20
25
5
10
15
20
25
10
15
20
25
Other
Summer
5
J-Mast wind speed [m/s]
Figure 7.6: Regression analysis of mast and SODAR wind speeds with low quality (quality index < 40) measurements highlighted in red, sites I and J.
65
CHAPTER 7. APPENDICES
7.2
Mast-SODAR regression analysis, part 2
A (100 m height)
A (100 m height)
20
15
10
5
00
5
10
15
A-Mast wind speed [m/s]
y = 1.02x + 0.83
y = 1.04x + 0.55
Unfiltered
Filtered
A-SODAR2 wind speed [m/s]
A-SODAR1 wind speed [m/s]
20
15
10
5
00
20
5
A (100 m height)
5
5
10
15
A-Mast wind speed [m/s]
y = 1.02x + 0.34
y = 1.02x + 0.34
Unfiltered
Filtered
B-SODAR1 wind speed [m/s]
A-SODAR3 wind speed [m/s]
10
15
10
5
00
20
5
C (100 m height)
10
15
B-Mast wind speed [m/s]
y = 1.00x + 0.33
y = 1.00x + 0.34
Unfiltered
Filtered
20
D (100 m height)
20
15
10
5
5
10
15
C-Mast wind speed [m/s]
y = 0.83x + 1.60
y = 1.02x + 0.50
Unfiltered
Filtered
D-SODAR wind speed [m/s]
20
C-SODAR wind speed [m/s]
20
20
15
00
15
B (120 m height)
20
00
10
A-Mast wind speed [m/s]
y = 0.97x + 0.46
y = 0.97x + 0.46
Unfiltered
Filtered
15
10
5
00
20
5
10
15
D-Mast2 wind speed [m/s]
y = 0.88x + 0.18
y = 0.94x - 0.27
Unfiltered
Filtered
20
Figure 7.7: The effect of quality control on mast-SODAR comparison using linear regression analysis, sites A-D. Blue values are original data and red values quality controlled data.
66
CHAPTER 7. APPENDICES
F (100 m height)
G (100 m height)
20
15
10
5
00
5
10
15
F-Mast3 wind speed [m/s]
y = 0.92x + 0.65
y = 0.92x + 0.63
Unfiltered
Filtered
G-SODAR wind speed [m/s]
F-SODAR wind speed [m/s]
20
15
10
5
00
20
5
H (100 m height)
5
5
10
15
H-Mast1 wind speed [m/s]
y = 0.88x + 0.08
y = 0.89x + 0.02
Unfiltered
Filtered
H-SODAR2 wind speed [m/s]
H-SODAR1 wind speed [m/s]
10
15
10
5
00
20
5
I (50 m height)
10
15
H-Mast2 wind speed [m/s]
y = 0.94x + 0.34
y = 0.98x + 0.10
Unfiltered
Filtered
20
J (100 m height)
20
15
10
5
5
10
15
I-Mast wind speed [m/s]
y = 0.82x + 0.72
y = 0.90x + 0.21
Unfiltered
Filtered
J-SODAR wind speed [m/s]
20
I-SODAR wind speed [m/s]
20
20
15
00
15
H (100 m height)
20
00
10
G-Mast wind speed [m/s]
y = 0.84x + 1.08
y = 0.90x + 0.71
Unfiltered
Filtered
15
10
5
00
20
5
10
15
J-Mast wind speed [m/s]
y = 0.96x + 0.55
y = 0.99x + 0.40
Unfiltered
Filtered
20
Figure 7.8: The effect of quality control on mast-SODAR comparison using linear regression analysis, sites F-J. Blue values are original data and red values quality controlled data.
67
CHAPTER 7. APPENDICES
7.3
60
B-Mast
60
Delta T
Sigma theta
50
40
Occurrence [%]
Occurrence [%]
50
Stability estimation methods
30
20
10
0 A
60
Occurrence [%]
50
K-Mast
Delta T
Sigma theta
40
30
20
10
B
D
E
C
Stability category
F
G
F
G
0 A
B
D
E
C
Stability category
F
G
L-Mast
Delta T
Sigma theta
40
30
20
10
0 A
B
D
E
C
Stability category
Figure 7.9: Stability categories using the sigma theta and delta T methods, sites B, K
and L.
68
CHAPTER 7. APPENDICES
7.4
Dependency of wind speed difference on SODAR
5
0
5
0
5
15
10
A-SODAR1 wind speed [m/s]
20
A (100 m height, N = 18013)
Median
5
0
5
0
5
15
10
A-SODAR3 wind speed [m/s]
20
C (100 m height, N = 19218)
Median
5
0
5
0
5
15
10
C-SODAR wind speed [m/s]
20
Wind speed difference (A-Mast - A-SODAR2) [m/s]
Median
Wind speed difference (B-Mast - B-SODAR1) [m/s]
A (100 m height, N = 20527)
Wind speed difference (D-Mast2 - D-SODAR) [m/s]
Wind speed difference (C-Mast - C-SODAR) [m/s]
Wind speed difference (A-Mast - A-SODAR3) [m/s]
Wind speed difference (A-Mast - A-SODAR1) [m/s]
wind speed
A (100 m height, N = 13847)
Median
5
0
5
0
5
15
10
A-SODAR2 wind speed [m/s]
20
B (120 m height, N = 12124)
Median
5
0
5
0
5
15
10
B-SODAR1 wind speed [m/s]
20
D (100 m height, N = 44913)
Median
5
0
5
0
5
15
10
D-SODAR wind speed [m/s]
20
Figure 7.10: Mast-SODAR wind speed difference as a function of SODAR wind speed,
sites A-D.
69
Median
5
0
5
0
5
15
10
F-SODAR wind speed [m/s]
20
H (100 m height, N = 14284)
Median
5
0
5
0
5
15
10
H-SODAR2 wind speed [m/s]
20
Wind speed difference (G-Mast - G-SODAR) [m/s]
F (100 m height, N = 7814)
Wind speed difference (H-Mast1 - H-SODAR1) [m/s]
Wind speed difference (H-Mast2 - H-SODAR2) [m/s]
Wind speed difference (F-Mast3 - F-SODAR) [m/s]
CHAPTER 7. APPENDICES
G (100 m height, N = 15783)
Median
5
0
5
0
5
Median
5
0
5
0
5
0
5
Wind speed difference (J-Mast - J-SODAR) [m/s]
Wind speed difference (I-Mast - I-SODAR) [m/s]
5
15
10
I-SODAR wind speed [m/s]
15
10
H-SODAR1 wind speed [m/s]
20
J (100 m height, N = 2870)
Median
5
20
H (100 m height, N = 12876)
I (50 m height, N = 33801)
0
15
10
G-SODAR wind speed [m/s]
20
Median
5
0
5
0
5
15
10
J-SODAR wind speed [m/s]
20
Figure 7.11: Mast-SODAR wind speed difference as a function of SODAR wind speed,
sites F-J.
70
CHAPTER 7. APPENDICES
7.5
Dependency of wind speed difference on tur-
5
0
5
0
10
20
30
40
50
A-Mast turbulence intensity [%]
60
70
A (100 m height, N = 18013)
Median
5
0
5
0
10
20
30
40
50
A-Mast turbulence intensity [%]
60
70
C (100 m height, N = 19218)
Median
5
0
5
0
10
20
30
40
50
C-Mast turbulence intensity [%]
60
70
Wind speed difference (A-Mast - A-SODAR2) [m/s]
Median
Wind speed difference (B-Mast - B-SODAR1) [m/s]
A (100 m height, N = 20527)
Wind speed difference (D-Mast2 - D-SODAR) [m/s]
Wind speed difference (C-Mast - C-SODAR) [m/s]
Wind speed difference (A-Mast - A-SODAR3) [m/s]
Wind speed difference (A-Mast - A-SODAR1) [m/s]
bulence intensity
A (100 m height, N = 13847)
Median
5
0
5
0
10
20
30
40
50
A-Mast turbulence intensity [%]
60
70
B (120 m height, N = 12124)
Median
5
0
5
0
10
20
30
40
50
B-Mast turbulence intensity [%]
60
70
D (100 m height, N = 44913)
Median
5
0
5
0
10
20
30
40
50
D-Mast2 turbulence intensity [%]
60
70
Figure 7.12: Mast-SODAR wind speed difference as a function of turbulence intensity,
sites A-D.
71
Median
5
0
5
0
10
20
30
40
50
F-Mast3 turbulence intensity [%]
60
70
H (100 m height, N = 12876)
Median
5
0
5
0
10
20
30
40
50
H-Mast1 turbulence intensity [%]
60
70
Wind speed difference (G-Mast - G-SODAR) [m/s]
F (100 m height, N = 7814)
Wind speed difference (H-Mast2 - H-SODAR2) [m/s]
Wind speed difference (H-Mast1 - H-SODAR1) [m/s]
Wind speed difference (F-Mast3 - F-SODAR) [m/s]
CHAPTER 7. APPENDICES
G (100 m height, N = 15783)
Median
5
0
5
0
10
Wind speed difference (J-Mast - J-SODAR) [m/s]
Wind speed difference (I-Mast - I-SODAR) [m/s]
5
0
5
20
30
40
50
I-Mast turbulence intensity [%]
70
Median
5
0
5
0
10
20
30
40
50
H-Mast2 turbulence intensity [%]
60
70
J (100 m height, N = 2870)
Median
10
60
H (100 m height, N = 14284)
I (50 m height, N = 33801)
0
20
30
40
50
G-Mast turbulence intensity [%]
60
70
Median
5
0
5
0
10
20
30
40
50
J-Mast turbulence intensity [%]
60
70
Figure 7.13: Mast-SODAR wind speed difference as a function of turbulence intensity,
sites F-J.
72
CHAPTER 7. APPENDICES
7.6
Dependency of wind speed difference on re-
Wind speed difference (A-Mast - A-SODAR3) [m/s]
A (100 m height, N = 20527)
Median
5
0
5
20
30
40
50
60
70
80
A-Mast relative humidity [%]
90
Wind speed difference (A-Mast - A-SODAR2) [m/s]
Wind speed difference (A-Mast - A-SODAR1) [m/s]
lative humidity
100
A (100 m height, N = 13847)
Median
5
0
5
20
30
40
50
60
70
80
A-Mast relative humidity [%]
90
100
A (100 m height, N = 18013)
Median
5
0
5
20
30
40
50
60
70
80
A-Mast relative humidity [%]
90
100
Figure 7.14: Mast-SODAR wind speed difference as a function of relative humidity for
site A, SODARs 1-3.
73
CHAPTER 7. APPENDICES
7.7
Dependency of wind shear on stability category
D
E
Alpha
F
C-Mast
N
1.0
20
0.5
15
0.0
10
0.5
5
1.0A
B
1.5
C
D
E
Stability category
Alpha
D-Mast1
N
1.0
10
8
0.5
6
0.0
4
0.5
1.0A
2
B
C
D
E
Stability category
F
Sample size (thousands)
4
0.5
2
B
C
D
E
Stability category
Alpha
G0
F
D-Mast2
N
1.0
20
15
0.5
10
0.0
5
0.5
B
1.5
G0
10
6
0.0
1.0A
12
12
8
0.5
1.5
G0
F
1.0
1.0A
25
N
Sample size (thousands)
Alpha (Mast, 70 m - 100 m)
1.5
C
Stability category
Alpha (Mast, 70 m - 100 m)
B
Alpha
B-Mast
C
D
E
Stability category
Alpha
G0
F
E-Mast
N
25
Sample size (thousands)
0.5
Alpha (Mast, 82 m - 120 m)
0.0
1.5
Alpha (Mast, 70 m - 100 m)
0.5
Sample size (thousands)
1.0
1.0A
Alpha (Mast, 70 m - 100 m)
N
45
40
35
30
25
20
15
10
5
G0
Sample size (thousands)
Alpha
A-Mast
Sample size (thousands)
Alpha (Mast, 70 m - 100 m)
1.5
1.0
20
0.5
15
0.0
10
0.5
5
1.0A
B
C
D
E
Stability category
F
G0
Figure 7.15: Average alpha value in each stability category, sites A-E. The red line
denotes number of measurements for each category.
74
CHAPTER 7. APPENDICES
0.0
10
0.5
5
E
Alpha
G0
F
F-Mast1
N
20
0.5
15
0.0
10
0.5
5
1.0A
B
1.5
C
D
E
Stability category
Alpha
G0
F
H-Mast2
N
1.0
25
0.5
20
0.0
15
10
0.5
1.0A
5
B
C
D
E
Stability category
F
0.5
15
0.0
10
0.5
5
B
Alpha
D
E
G0
F
G-Mast
N
35
30
25
0.5
20
0.0
15
10
0.5
5
B
1.5
G0
C
Stability category
1.0
1.0A
35
30
20
1.5
25
1.0
1.0
1.0A
Alpha (Mast, 70 m - 100 m)
D
Alpha (Mast, 70 m - 100 m)
Alpha (Mast, 57 m - 70 m)
1.5
C
Stability category
Sample size (thousands)
B
25
Sample size (thousands)
15
N
Sample size (thousands)
0.5
Alpha
F-Mast2
C
D
E
Stability category
Alpha
H-Mast1
N
1.0
0.5
0.0
0.5
1.0A
B
C
D
G0
F
E
Stability category
F
45
40
35
30
25
20
15
10
5
G0
Sample size (thousands)
20
Alpha (Mast, 57 m - 70 m)
1.0
1.0A
Alpha (Mast, 70 m - 100 m)
N
1.5
25
Sample size (thousands)
Alpha
F-Mast3
Sample size (thousands)
Alpha (Mast, 70 m - 100 m)
1.5
Figure 7.16: Average alpha value in each stability category, sites F-H. The red line
denotes number of measurements for each category.
75
CHAPTER 7. APPENDICES
B
1.5
C
D
E
Stability category
Alpha
F
L-Mast
N
1.0
35
30
25
0.5
20
0.0
15
10
0.5
1.0A
5
B
C
D
E
Stability category
F
1.0
25
15
0.0
10
0.5
5
B
1.5
G0
30
20
0.5
1.0A
40
N
Sample size (thousands)
0.5
Alpha
K-Mast
C
D
E
Stability category
Alpha
G0
F
M-Mast
N
1.0
20
Sample size (thousands)
0.0
Alpha (Mast, 58 m - 82 m)
0.5
1.5
Alpha (Mast, 80 m - 100 m)
1.0
1.0A
Alpha (Mast, 82 m - 120 m)
N
45
40
35
30
25
20
15
10
5
G0
Sample size (thousands)
Alpha
J-Mast
Sample size (thousands)
Alpha (Mast, 70 m - 100 m)
1.5
15
0.5
10
0.0
5
0.5
1.0A
B
C
D
E
Stability category
F
G0
Figure 7.17: Average alpha value in each stability category, sites J-M. The red line
denotes number of measurements for each category.
76
CHAPTER 7. APPENDICES
7.8
Seasonal wind profiles
A
Other
120
100
100
100
80
60
A-Mast
A-SODAR1
A-SODAR2
A-SODAR3
203
4
5
6
7
Wind speed [m/s]
8
Height [m]
120
40
80
60
A-Mast
A-SODAR1
A-SODAR2
A-SODAR3
40
9
203
4
5
6
7
Wind speed [m/s]
8
80
60
9
203
100
100
100
80
60
B-Mast
B-SODAR1
B-SODAR2
4
5
6
7
Wind speed [m/s]
8
Height [m]
120
Height [m]
120
203
4
5
80
60
B-Mast
B-SODAR1
B-SODAR2
40
9
203
4
5
6
7
Wind speed [m/s]
8
9
8
80
60
B-Mast
B-SODAR1
B-SODAR2
40
9
203
4
5
6
7
Wind speed [m/s]
Figure 7.18: Seasonal mean wind profiles, sites A and B.
77
7
6
Wind speed [m/s]
Summer
120
40
A-Mast
A-SODAR1
A-SODAR2
A-SODAR3
40
B
Other
Winter
Height [m]
Summer
120
Height [m]
Height [m]
Winter
8
9
CHAPTER 7. APPENDICES
C
Other
120
100
100
100
80
60
203
5
6
7
Wind speed [m/s]
8
80
60
40
C-Mast
C-SODAR
4
Height [m]
120
40
9
203
4
5
6
7
Wind speed [m/s]
8
80
60
40
C-Mast
C-SODAR
9
203
100
100
100
80
60
D-Mast2
D-Mast1
D-SODAR
4
5
6
7
Wind speed [m/s]
8
Height [m]
120
Height [m]
120
203
4
5
80
60
D-Mast2
D-Mast1
D-SODAR
40
9
203
4
5
6
6
7
Wind speed [m/s]
8
9
8
80
60
D-Mast2
D-Mast1
D-SODAR
40
9
203
4
5
6
7
Wind speed [m/s]
Figure 7.19: Seasonal mean wind profiles, sites C and D.
78
7
Wind speed [m/s]
Summer
120
40
C-Mast
C-SODAR
D
Other
Winter
Height [m]
Summer
120
Height [m]
Height [m]
Winter
8
9
CHAPTER 7. APPENDICES
E
Other
120
100
100
100
80
60
203
5
6
7
Wind speed [m/s]
8
80
60
40
E-Mast
E-SODAR
4
Height [m]
120
40
9
203
4
5
6
7
Wind speed [m/s]
8
80
60
40
E-Mast
E-SODAR
9
203
100
100
100
80
60
F-Mast1
F-Mast2
F-Mast3
F-SODAR
4
5
6
7
Wind speed [m/s]
8
Height [m]
120
Height [m]
120
203
4
5
80
60
F-Mast1
F-Mast2
F-Mast3
F-SODAR
40
9
203
4
5
6
6
7
Wind speed [m/s]
8
9
8
80
60
F-Mast1
F-Mast2
F-Mast3
F-SODAR
40
9
203
4
5
6
7
Wind speed [m/s]
Figure 7.20: Seasonal mean wind profiles, sites E and F.
79
7
Wind speed [m/s]
Summer
120
40
E-Mast
E-SODAR
F
Other
Winter
Height [m]
Summer
120
Height [m]
Height [m]
Winter
8
9
CHAPTER 7. APPENDICES
G
Other
120
100
100
100
80
60
203
5
6
7
Wind speed [m/s]
8
80
60
40
G-Mast
G-SODAR
4
Height [m]
120
40
9
203
4
5
6
7
Wind speed [m/s]
8
80
60
40
G-Mast
G-SODAR
9
203
100
100
100
80
60
H-Mast2
H-Mast1
H-SODAR2
H-SODAR1
4
5
6
7
Wind speed [m/s]
8
Height [m]
120
Height [m]
120
203
4
5
80
60
H-Mast2
H-Mast1
H-SODAR2
H-SODAR1
40
9
203
4
5
6
7
Wind speed [m/s]
8
9
8
80
60
H-Mast2
H-Mast1
H-SODAR2
H-SODAR1
40
9
203
4
5
6
7
Wind speed [m/s]
Figure 7.21: Seasonal mean wind profiles, sites G and H.
80
7
6
Wind speed [m/s]
Summer
120
40
G-Mast
G-SODAR
H
Other
Winter
Height [m]
Summer
120
Height [m]
Height [m]
Winter
8
9
CHAPTER 7. APPENDICES
I
Other
120
100
100
100
80
60
203
5
6
7
Wind speed [m/s]
8
80
60
40
I-Mast
I-SODAR
4
Height [m]
120
40
9
203
4
5
6
7
Wind speed [m/s]
8
80
60
40
I-Mast
I-SODAR
9
203
100
100
100
80
60
5
6
7
Wind speed [m/s]
8
80
60
40
J-Mast
J-SODAR
4
Height [m]
120
Height [m]
120
203
4
5
9
203
5
6
7
Wind speed [m/s]
8
9
8
60
9
203
J-Mast
J-SODAR
4
5
6
7
Wind speed [m/s]
Figure 7.22: Seasonal mean wind profiles, sites I and J.
81
7
80
40
J-Mast
J-SODAR
4
6
Wind speed [m/s]
Summer
120
40
I-Mast
I-SODAR
J
Other
Winter
Height [m]
Summer
120
Height [m]
Height [m]
Winter
8
9
CHAPTER 7. APPENDICES
K
Other
120
100
100
100
80
60
203
Height [m]
120
40
80
60
40
K-Mast
4
5
6
7
Wind speed [m/s]
8
9
203
80
60
40
K-Mast
4
5
6
7
Wind speed [m/s]
8
9
203
100
100
100
80
60
40
Height [m]
120
Height [m]
120
80
60
40
L-Mast
4
5
6
5
7
Wind speed [m/s]
8
9
203
6
7
Wind speed [m/s]
8
9
Summer
120
203
K-Mast
4
L
Other
Winter
Height [m]
Summer
120
Height [m]
Height [m]
Winter
80
60
40
L-Mast
4
5
6
7
Wind speed [m/s]
8
9
203
L-Mast
4
5
7
Wind speed [m/s]
Figure 7.23: Seasonal mean wind profiles, sites K and L.
82
6
8
9
CHAPTER 7. APPENDICES
M
Other
120
100
100
100
80
60
203
Height [m]
120
40
80
60
40
M-Mast
4
5
6
7
Wind speed [m/s]
8
9
203
80
60
40
M-Mast
4
5
6
7
Wind speed [m/s]
8
9
203
100
100
100
80
60
40
Height [m]
120
Height [m]
120
80
60
40
N-SODAR
4
5
6
5
7
Wind speed [m/s]
8
9
203
6
7
Wind speed [m/s]
8
9
Summer
120
203
M-Mast
4
N
Other
Winter
Height [m]
Summer
120
Height [m]
Height [m]
Winter
80
60
40
N-SODAR
4
5
6
7
Wind speed [m/s]
8
9
203
N-SODAR
4
5
7
Wind speed [m/s]
Figure 7.24: Seasonal mean wind profiles, sites M and N.
83
6
8
9
CHAPTER 7. APPENDICES
O
Other
120
100
100
100
80
60
203
Height [m]
120
40
80
60
40
O-SODAR
4
5
6
7
Wind speed [m/s]
8
9
203
80
60
40
O-SODAR
4
5
6
7
Wind speed [m/s]
8
9
203
100
100
100
80
60
40
Height [m]
120
Height [m]
120
80
60
40
P-SODAR
4
5
6
5
7
Wind speed [m/s]
8
9
203
6
7
Wind speed [m/s]
8
9
Summer
120
203
O-SODAR
4
P
Other
Winter
Height [m]
Summer
120
Height [m]
Height [m]
Winter
80
60
40
P-SODAR
4
5
6
7
Wind speed [m/s]
8
9
203
P-SODAR
4
5
7
Wind speed [m/s]
Figure 7.25: Seasonal mean wind profiles, sites O and P.
84
6
8
9
CHAPTER 7. APPENDICES
Q
Other
Summer
120
120
100
100
100
80
60
40
203
Height [m]
120
Height [m]
Height [m]
Winter
80
60
40
Q-SODAR
4
5
6
7
Wind speed [m/s]
8
9
203
80
60
40
Q-SODAR
4
5
6
7
Wind speed [m/s]
8
9
203
Q-SODAR
4
6
7
Wind speed [m/s]
Figure 7.26: Seasonal mean wind profiles, site Q.
85
5
8
9
Bibliography
Proposed Revision 1 to Regulatory Guide 1.23. Meteorological Programs in Support of Nuclear Power Plants. Nuclear Regulatory Commission (NCR), 9 1980.
A. Anjan, M. Ranalkar, B. Amudha, R. Pratap, and R. D. Vashistha. Comparison of Wind data of Ultrasonic wind sensor with Dyne Tube Pressure Anemograph wind sensor and its performances in Indian Subcontinent, 2013. URL http://www.wmo.int/pages/prog/www/IMOP/publications/
IOM-96_TECO-2008/2(15)_Anjan_India.pdf,(Oct23rd2013).
AQSystem. AQ500 Wind Finder technical specifications. URL http://www.aqs.
se/solutions/technical-specifications/,(Oct23rd2013).
AQSystem. Personal communication, 2013.
AQSystem, 2013. URL http://www.aqs.se,(Oct23rd2013).
Atmospheric stability classification. Air Pollution Research Group, University of
Toledo, 2013. URL http://www.eng.utoledo.edu/aprg/courses/iap/text/
met/9_atmos_classify.html,(Oct23rd2013).
F. Bañuelos-Ruedas, C. Angeles-Camacho, and S. Rios-Marcuello. Analysis and
validation of the methodology used in the extrapolation of wind speed data
at different heights. Renewable and Sustainable Energy Reviews, 14(8):2383 –
2391, 2010. ISSN 1364-0321. doi: 10.1016/j.rser.2010.05.001.
S. Bradley. Wind speed errors for lidars and sodars in complex terrain. IOP
Conference Series: Earth and Environmental Science, 1(1):012061, 2008. doi:
10.1088/1755-1315/1/1/012061.
86
BIBLIOGRAPHY
S. Bradley. Atmospheric Acoustic Remote Sensing: Principles and Applications.
CRC Press, Taylor & Francis Group, 2010. ISBN 978-0-8493-3588-4.
S. Bradley, I. Antoniou, S. von Hünerbein, D. Kindler, M. de Noord, and H. Jörgensen. SODAR calibration for wind energy applications, final reporting on
WP3, EU WISE project NNE5-2001-297. Greater Manchester, UK, 3 2005.
ISBN 0-9541649-1-1.
S. Bradley, Y. Perrott, P. Behrens, and A. Oldroyd. Corrections for Wind-Speed
Errors from Sodar and Lidar in Complex Terrain. Boundary-Layer Meteorology,
143(1):37–48, 2012. ISSN 0006-8314. doi: 10.1007/s10546-012-9702-0.
F. V. Brock and S. J. Richardson. Meteorological Measurement Systems. Oxford
University Press, Inc., 2001. ISBN 0-19-513451-6.
T. Burton, D. Sharpe, N. Jenkins, and E. Bossanyi. Wind Energy Handbook.
Wiley, 2008. ISBN 0-471-48997-2.
A. Cenedese, G. Cosemans, H. Erbrink, and R. Stubi. Vertical profiles of wind,
temperature and turbulence. Harmonisation of the pre-processing of meteorological data for atmospheric dispersion models. COST action, 710, 1998.
A. Clifton, D. Elliott, and M. Courtney. 15. Ground-based Vertically-profiling
Remote Sensing for Wind Resource Assessment. Expert group study on recommended practises. International Energy Agency (IEA), 1 2013.
G. H. Crescenti. A look Back on Two Decades of Doppler Sodar Comparison
Studies. Bulletin of the American Meteorological Society, 78(4):651–673, 1997.
ISSN 0003-0007. doi: 10.1175/1520-0477(1997)078<0651:ALBOTD>2.0.CO;2.
G. H. Crescenti. The degradation of doppler sodar performance due to noise: a
review. Atmospheric Environment, 32(9):1499 – 1509, 1998. ISSN 1352-2310.
doi: 10.1016/S1352-2310(97)00385-3.
Delphi234.
Wind turbine size increase 1980-2011, 2012.
URL http:
//commons.wikimedia.org/wiki/File:Wind_turbine_size_increase_
1980-2011.png,(Oct23rd2013).
87
BIBLIOGRAPHY
N. R. Draper, H. Smith, and E. Pownell. Applied regression analysis, volume 3.
Wiley New York, 1966. ISBN ISBN 0-471-17082-8.
S. FAS.
Sodar principle.
URL http://sasakihome.info/~sasaki/
TibetSiteSurvey/Scintec_Sodar/SODAR_Principle.gif,(Oct23rd2013).
Finnish Meteorological Institute (FMI) and Technical Research Centre of Finland
(VTT). Finnish Icing Atlas. URL http://www.tuuliatlas.fi/icingatlas/
index.html,(Oct23rd2013).
E. Firtin, O. Güler, and S. A. Akdag. Investigation of wind shear coefficients
and their effect on electrical energy generation. Applied Energy, 88(11):4097 –
4105, 2011. ISSN 0306-2619. doi: 10.1016/j.apenergy.2011.05.025.
T. Foken. 50 years of the Monin–Obukhov Similarity Theory. BoundaryLayer Meteorology, 119(3):431–447, 2006. ISSN 0006-8314. doi: 10.1007/
s10546-006-9048-6.
N. I. Fox. A tall tower study of missouri winds. Renewable Energy, 36(1):330 –
337, 2011. ISSN 0960-1481. doi: 10.1016/j.renene.2010.06.047.
G. Gualtieri and S. Secci. Comparing methods to calculate atmospheric stabilitydependent wind speed profiles: A case study on coastal location. Renewable
Energy, 36(8):2189 – 2204, 2011. ISSN 0960-1481. doi: 10.1016/j.renene.2011.
01.023.
G. Gualtieri and S. Secci. Methods to extrapolate wind resource to the turbine
hub height based on power law: A 1-h wind speed vs. weibull distribution
extrapolation comparison. Renewable Energy, 43(0):183 – 200, 2012. ISSN
0960-1481. doi: 10.1016/j.renene.2011.12.022.
G. Gualtieri and S. Secci. Extrapolating wind speed time series vs. weibull distribution to assess wind resource to the turbine hub height: A case study on
coastal location in southern italy. Renewable Energy, 62(0):164 – 176, 2013.
ISSN 0960-1481. doi: 10.1016/j.renene.2013.07.003.
J. R. Holton. An Introduction to Dynamic Meteorology. Elsevier Academic Press,
2004. ISBN 0-12-354015-1.
88
BIBLIOGRAPHY
C. H. Hunter. A Recommended Pasquill-Gifford Stability Classification Method
for Safety Basis Atmospheric Dispersion Modeling at SRS, 3 2012.
International Electrotechinal Commission, 2005. IEC 61400-12-1, Wind turbines
- Part 12-1: Power performance measurements of electricity producing wind
turbines. International Electrotechinal Commission (IEC), Geneva, Switzerland, 1 edition, 12 2005.
J. Junge and A. Westerhellweg. Estimation of Zero Plane Displacement Heights
in the Vicinity of a Forest, 2013.
L. Kristensen. Cup anemometer behavior in turbulent environments. Journal
of Atmospheric and Oceanic Technology, 15:5–17, 2 1998. doi: 10.1175/
1520-0426(1998)015<0005:CABITE>2.0.CO;2.
M. A. Lackner, A. L. Rogers, J. F. Manwell, and J. G. McGowan. A new method
for improved hub height mean wind speed estimates using short-term hub
height data. Renewable Energy, 35(10):2340 – 2347, 2010. ISSN 0960-1481.
doi: 10.1016/j.renene.2010.03.031.
S. Lang and E. McKeogh. LIDAR and SODAR Measurements of Wind Speed
and Direction in Upland Terrain for Wind Energy Purposes. Remote Sensing,
3(10.3390/rs3091871):1871–1901, 8 2011. ISSN 2072-4292.
Long-Term Climate and Energy Strategy. Ministerial Working Group on Climate Change and Energy, Ministry of employment and the economy, 2013.
URL http://www.tem.fi/files/36292/Energia-_ja_ilmastostrategia_
nettijulkaisu_ENGLANNINKIELINEN.pdf,(Oct23rd2013).
K. E. Moore and B. H. Bailey. Maximizing the Accuracy of Sodar Measurements
for Wind Resource Assessment. AWS Truewind, 8 2006.
K. E. Moore and B. H. Bailey. Recommended Practises for the Use of Sodar in
Wind Energy Resource Assessment, 7 2011.
T. R. Oke. Boundary Layer Climates. Methuen, 2 edition, 1978.
89
BIBLIOGRAPHY
E. L. Petersen, N. G. Mortensen, L. Landberg, J. Højstrup, and H. P. Frank.
Wind power meteorology. part i: Climate and turbulence. Wind Energy, 1(S1):
2–22, 7 1998. ISSN 1095-4244. doi: 10.1002/(SICI)1099-1824(199809)1:1<2::
AID-WE15>3.0.CO;2-Y.
E. W. Peterson and J. P. Hennessey. On the use of power laws for estimates of
wind power potential. Journal of Applied Meteorology, 17:390–394, 1978. doi:
10.1175/1520-0450(1978)017<0390:OTUOPL>2.0.CO;2.
M. Ray, A. Rogers, and J. McGowan. Analysis of wind shear models and trends
in different terrains. Wind Engineering, 30(4):341–350, 2006. doi: 10.1260/
030952406779295417.
S. Rehman and N. M. Al-Abbadi. Wind shear coefficients and their effect on
energy production. Energy Conversion and Management, 46(15-16):2578–2591,
2005. ISSN 0196-8904. doi: 10.1016/j.enconman.2004.12.005.
A. Romo Perea, J. Amezcua, and O. Probst. Validation of three new measurecorrelate-predict models for the long-term prospection of the wind resource.
Journal of Renewable & Sustainable Energy, 3(2):023105, 2011. ISSN 19417012. doi: 10.1063/1.3574447.
J. Sanz Rodrigo, F. Borbón Guillén, P. Gómez Arranz, M. S. Courtney, R. Wagner, and E. Dupont. Multi-site testing and evaluation of remote sensing instruments for wind energy applications. Renewable Energy, 53:200–210, 2013.
ISSN 0960-1481. doi: 10.1016/j.renene.2012.11.020.
G. N. Scott, D. L. Elliott, and M. N. Schwartz. Comparison of second wind triton data with meteorological tower measurements. Technical report, National
Renewable Energy Laboratory, 2010.
K. Smith, G. Randall, D. Malcolm, N. Kelley, and B. Smith. Evaluation of wind
shear patterns at midwest wind energy facilities: Preprint. Technical report,
National Renewable Energy Lab., Golden, CO (US), 2002.
R. B. Stull. An Introduction to Boundary Layer Meteorology. Kluwer Academic
Publishers, 1988.
90
BIBLIOGRAPHY
Y. Tamura, Y. Iwatani, K. Hibi, K. Suda, O. Nakamura, T. Maruyama, and
R. Ishibashi. Profiles of mean wind speeds and vertical turbulence intensities
measured at seashore and two inland sites using doppler sodars. Journal of
Wind Engineering and Industrial Aerodynamics, 95(6):411–427, 2007. ISSN
0167-6105. doi: 10.1016/j.jweia.2006.08.005.
Thies Clima, 2013.
(Oct23rd2013).
URL http://www.thiesclima.com/wind_e.html,
J. S. Touma. Dependence of the wind profile power law on stability for various
locations. Journal of the Air Pollution Control Association, 27(9):863–866,
1977. doi: 10.1080/00022470.1977.10470503.
TuuliWatti Oy. Toivo aloittaa tuotannon Iin Olhavan tuulivoimapuistossa,
3 2013.
URL http://www.tuuliwatti.fi/index.php?id=12340,
(Oct23rd2013).
Wind Energy Statistics in Finland. VTT - Technical Research Centre of Finland,
2013. URL http://www.vtt.fi/proj/windenergystatistics/index.jsp?
lang=en,(Oct23rd2013).
J. C. Wyngaard. Cup, propeller, vane, and sonic anemometers in turbulence
research. Annual Review of Fluid Mechanics, 13(1):399–423, 1981. doi: 10.
1146/annurev.fl.13.010181.002151.
91