sodar verification test - vaisala · 2018. 7. 20. · the verification analysis is based on the iec...
Post on 28-Aug-2021
2 Views
Preview:
TRANSCRIPT
SoDAR Verification Test Triton SoDAR at Test Site Lelystad
05.03.2015– 29.05.2015
- Confidential –
ECOFYS WTTS B.V.
Kanaalweg 15G | 3526 KL Utrecht| T +31 (0)30 662-3827 | E info@ecofyswtts.com | I www.ecofyswtts.com
Chamber of Commerce 24468357
SoDAR Verification Test Triton SoDAR at Test Site Lelystad
05.03.2015– 29.05.2015
- Confidential –
Project number:
Author:
WIEWT15389
Federico Montealegre
14/07/2015
Reviewer: Anthony Crockford 21/07/2015
Approval: Erik Holtslag 22/07/2015
Filename 20151116_REP_Triton_SoDAR I_VER_MM01_Final
Pages 47
Version Date
Author
Checked
by
Approved
by
Remarks/Change
0.1 27/07/2015 FMo ACr EHo Draft for client review
1.0 03/12/2015 FMo ACr Eho Final
© Ecofys WTTS 2015 by order of: Vaisala Inc.
ECOFYS WTTS B.V.
Kanaalweg 15G | 3526 KL Utrecht| T +31 (0)30 662-3827 | E info@ecofyswtts.com | I www.ecofyswtts.com
Chamber of Commerce 24468357
Summary
On request of Vaisala Inc., Ecofys Wind Turbine Testing Services (WTTS) carried out a verification of a
Triton SoDAR (Triton SoDAR I, Serial number 604) against the reference met mast TSL-MM01 at the
WTTS Test Site Lelystad (TSL).
The 120 m TSL-MM01 is fully IEC and MEASNET compliant and equipped with high quality cup-
anemometers at 40, 80, 100 and 120 m. In addition, wind vanes for wind direction measurements are
located at 87 and 118 m. The SoDAR was collecting wind data at four common heights with the met
mast. The SoDAR verification campaign lasted from 05/03/2015 to 29/05/2015.
All measured data was collected at the specified location and filtered to ensure entirely valid datasets.
The verification analysis is based on the IEC standard 61400-12-1 (ed 2, draft [2]). The Triton SoDAR
604 verification test covers a full three months of collocated measurements during spring and summer
2015, and has permitted the collection of over 5,000 concurrent valid data points. This resulting dataset
is significantly larger than the minimum required, providing a robust basis for uncertainty calculations
in each of the 0.5m/s wind speed bins up to 16 m/s and demonstrates the SoDAR’s performance across
seasons.
Linear regression between the overall SoDAR and met mast recordings showed consistent, highly-
correlated measurements with slopes near unity. The calculated uncertainty in the SoDAR wind speed
measurements is low, in line with first-class anemometry for the majority of the wind speed bins. The
calculated uncertainty tables (in Appendix C) can be used directly in wind resource assessments,
together with the classification uncertainty and site-specific uncertainty components.
Sensitivity tests of the wind speed deviation revealed that the wind speed deviation shows no significant
linear correlation to external conditions: vertical wind speed, horizontal wind speed, turbulence
intensity or rain. This indicates that the SoDAR is insensitive to these factors.
The SODAR measurements were also validated against NORSEWiND criteria for LiDAR remote sensors
[3] and meets the Ecofys WTTS acceptance thresholds for field measurements. There is scatter in the
individual recordings, which is partly due to the distance between the SoDAR and mast during the test.
The overall high correlations and very good linear regression fit indicate that the SoDAR is functioning
properly with high accuracy.
As a result of this test, Ecofys WTTS judges that the SoDAR Triton 604 is suitable for field
measurements. The SODAR will measure the long-term mean wind speeds with accuracy comparable
to cup anemometry in flat terrain. However, when analysing Triton data sets, due consideration should
be made for the standard deviation of 10-minute values characteristic to the system.
ECOFYS WTTS B.V.
Kanaalweg 15G | 3526 KL Utrecht| T +31 (0)30 662-3827 | E info@ecofyswtts.com | I www.ecofyswtts.com
Chamber of Commerce 24468357
Table of contents
1 Introduction 5
1.1 Scope of the study 5
1.2 IEC 61400-12-1 (ed 2) 5
1.3 NORSEWInD verification criteria 6
1.4 Structure of the Report 7
2 Verification Campaign 8
2.1 Site description 8
2.2 Reference IEC-compliant Met Mast description “MM01” 9
2.3 SoDAR Location 10
2.4 SoDAR Orientation and Time Synchronization 11
2.5 Valid wind direction sectors 12
3 SoDAR Verification Procedure 15
3.1 Met Mast Data Filtering 15
3.2 SoDAR Data Filtering 15
3.3 Statistical tests 16
4 Results & Discussion 19
4.1 Data coverage 19
4.2 Wind speed verification 21
4.3 Wind direction verification 23
4.4 Sensitivity tests 26
4.5 Verification uncertainty analysis 27
4.6 NORSEWInD criteria validation 30
5 Conclusions 31
Appendix A Anemometer calibration certificates 33
Appendix B Uncertainty analysis 37
Appendix C Detailed uncertainty analysis 41
WIEWT15389 – Triton SoDAR 131676 Verification 5
1 Introduction
Ecofys WTTS verified the operation of the Triton SoDAR with serial number 604, prior to its application
in a wind measurement campaign. The Triton SoDAR 604 is verified against a fully IEC- and MEASNET-
compliant met mast (120 m) at the Ecofys Test Site Lelystad (TSL). A complete and detailed
assessment of the SoDAR verification process is presented with special attention to sensitivity tests
and uncertainty.
1.1 Scope of the study
This study verifies the accuracy of the Triton SoDAR with serial number 604, and determines whether
it operates as specified by the manufacturer to ensure that the SoDAR measurements from this unit
are traceable to international standards for use in wind resource assessments. The verification
procedure evaluates the accuracy of the SODAR measurements based on two international wind
industry standards:
IEC 61400-12-1 verification procedure and uncertainty evaluation [1] [2]
NORSEWInD SoDAR validation criteria [3]
1.2 IEC 61400-12-1 (ed 2)
IEC 61400-12-1 (ed 1 [1]) is the definitive industry-wide standard for high-quality wind measurement
campaigns using standard anemometry. The second edition (IEC 61400-12-1 (ed 2)*) also specifies
the use of SoDAR, with a detailed procedure (Annex L) that ensures the traceability of the
measurements and evaluates associated uncertainty components, which can be applied in wind
resource assessments:
“This test is a comparison of the remote sensing device measurements to those from calibrated cup
anemometers mounted on a mast spanning a significant portion of the height range of interest. The
purpose of this test is to convey traceability to international standards to this particular device, in the
form of an uncertainty. A second result of the verification test is an assessment of the random noise of
the device.” [2]
The systematic uncertainties in the SoDAR measurements are evaluated for each 0.5 m/s wind speed
bin from 4-16 m/s:
“The standard uncertainty of the reference sensor”;
“The mean deviation of the remote sensing device measurements and the reference sensor
measurements”;
“The standard deviation of the measurement of the RS device calculated as the standard
deviation of the measurements divided by the square root of the number of data per bin”;
“Uncertainty of the remote sensing device due to mounting effects”; and
“Uncertainty of the remote sensing device due to non-homogenous flow.” [2]
* Since the second edition is not yet published, the analysis is based on a Committee Draft [2]
WIEWT15389 – Triton SoDAR 131676 Verification 6
The wind speed dataset should include at least 3 pairs of valid measurements in each wind speed bin
between 4 m/s and 16 m/s, and the total amount of valid data should be minimum 180 hours. For
practical reasons, Ecofys WTTS normally limits the verification campaign to a period of 8 weeks, which
should be sufficient to meet these criteria.
Wind direction measurements are validated in 5 degree bins by means of regression analysis between
SoDAR and met mast wind vane measurements.
1.3 NORSEWInD verification criteria
The IEC procedure does not establish thresholds for the accuracy of the SoDAR measurements; for this
reason, Ecofys WTTS also applies the validation criteria from the EU NORSEWInD project. Even though
these criteria were designed for LiDAR, it is used in this analysis as a reference, as these criteria are
used throughout the wind industry. These criteria can also be evaluated with a shorter campaign so
the SoDAR can be re-validated prior to each measurement campaign – with a comparison to these
results.
NORSEWInD has defined validation criteria to evaluate the absolute error and the quality of the linear
regression between Remote Sensors and anemometry:
“Absolute error – difference in reported wind speed between the reference and test instrument
based on 10-minute averages”;
“Linear regression gradient – this is based on a single variant regression, with the regression
analysis constrained to pass through origin (y=mx+b; b==0)”; and
“Linear regression R-squared values – is the quality of fit value returned from the analysis
performed to assess the linear regression gradient value.” [3]
The NORSEWInD criteria recommends at least 600 valid data points. As noted above, Ecofys WTTS
limited the verification campaign to a period of 12 weeks max., which was sufficient to gather these
600 valid data points. If necessary, the analysis may be based on a smaller sample; to maintain an
accurate comparison, a minimum of 200 valid data points is needed.
The validation criteria are shown in Table 1. The primary comparison will be between measurements
at the top measurement height of the met mast.
If sufficient valid data is collected within the measurement period, the data will also be compared in
two wind speed ranges. This can help to identify non-linearities.
Table 1: NORSEWInD Criteria [2]
Criteria NORSEWInD Threshold
Absolute error <0.5m/s Not more than 10% of data to exceed this value
Linear regression slope Between 0.98 and 1.01
Linear regression R² >0.98
WIEWT15389 – Triton SoDAR 131676 Verification 7
If the SODAR is compliant with the NORSEWInD thresholds, Ecofys WTTS will deem that the unit can
be used for field measurements as a replacement to a met mast, while maintaining a similar level of
accuracy to IEC-compliant cup anemometry.
Ecofys WTTS recognises that there can be a wide range of other purposes for wind measurements with
remote-sensing devices, and also that the NORSEWInD criteria were designed for LiDAR evaluations,
where the device can be located directly at the mast base, which is not possible for SoDARs. Thus,
Ecofys WTTS considers modified thresholds to determine if the unit is suitable in general for field
measurements. These thresholds consider only the overall linear regression and allow for a slightly
wider tolerance. The threshold levels are shown in Table 2. The primary comparison will be between
measurements at the top measurement height of the met mast.
Table 2: Ecofys WTTS Acceptance Thresholds, based on NORSEWInD criteria [2]
Criterion Category
NORSEWInD Acceptance
Threshold for Replacement of IEC-compliant Cup Anemomery
Ecofys WTTS
Acceptance Threshold for Field Measurements
Number of valid data points
4-8m/s >200
8-12 m/s >200
ALL >600
Percentage of data points that exceed 0.5 m/s absolute error
ALL <10% n/a
Linear regression slope
4-8m/s 0.98-1.01 n/a
8-12 m/s 0.98-1.01 n/a
Variation in slope
<0.015 n/a
ALL 0.98-1.01 0.98-1.02
Linear regression – R2
4-8m/s >0.98 n/a
8-12 m/s >0.98 n/a
ALL >0.98 >0.95
1.4 Structure of the Report
First, Test Site Lelystad is described in detail including its surroundings, wind turbines, met masts,
obstacles and orography. Moreover, the SoDAR verification process is designed based on the met mast
layout, instrumentation, valid wind sectors and time synchronisation.
Subsequently, the verification procedure is described. Data filtering and data quality are defined and
summarised and statistical methods are explained. Results for wind speed and wind direction are
presented in the ‘Results and discussion’ chapter including a small discussion subsection per variable.
This chapter also covers on data filtering, NORSEWInD verification criteria and Ecofys WTTS’ SoDAR
thresholds, a sensitivity analysis and an uncertainty analysis. Finally, conclusions are drawn.
WIEWT15389 – Triton SoDAR 131676 Verification 8
2 Verification Campaign
2.1 Site description
Test site Lelystad is located in a flat, open landscape in the centre of the Netherlands with meadows,
scattered houses and several wind turbines nearby. Within a 4 km radius of the site, the terrain does
not vary more than 1.6 m (from 2.9 to 4.5 m below sea level [4]). Including the water level in the
canals, which is about 2 m lower than surrounding land, the slope in the region never exceeds 2% and
all positions at Test Site Lelystad are fully compliant with the IEC 61400-12-1 Terrain Assessment [1].
Figure 1: Terrain at TSL. TSL-MM01 is located to the northwest of wind turbine WTG01 (it is shown as a red triangle).
WIEWT15389 – Triton SoDAR 131676 Verification 9
2.2 Reference IEC-compliant Met Mast description “MM01”
The verification is carried out against the fully IEC compliant instruments on the TSL-MM01 met mast,
whose location is specified in Table 3.
Table 3: Coordinates of met mast TSL-MM01
Easting [RD] Northing [RD]
Met mast TSL-MM01 168,522 504,345
The instrumentation used for this verification is specified in Table 4.
Table 4: Partial sensor list of met mast TSL-MM01
Sensor Height Location Position
(orientation) Sensor type
Serial
Number*
DKD
Calibration
number
Anemometer A1
120 Top
anemometer N/A
Thies TFC adv.
0609242 1436778
Anemometer A2
100 Double boom 349° Thies TFC
adv. 0609153 1436791
Anemometer A3
80 Double boom 169° Thies TFC
adv. 0408875 1436775
Anemometer A4
40 Single boom 169° Thies TFC
adv. 0408870 1436774
Wind Vane WF1
118 Single boom 351° Thies TFC 06131639 /
Wind Vane WF2
40 Single boom 349° Thies TFC 08141535 /
One top anemometer is installed at 120 m next to a lighting protection rod. The lightning rod is about
5 m tall at a distance of 1.3 m from the top-anemometer, as shown in the drawing below. The support
boom of the lightning rod is oriented towards 150°. This design of the lightning detector ensures limited
flow distortion of the top-anemometer recordings according to specifications in [1].
Figure 2: Top-section of TSL-MM01 with top anemometer, wind vane and the lightning rod.
* Serial numbers are used as reference to the calibration certificates from Deutsche Windguard, shown in Appendix A.
WIEWT15389 – Triton SoDAR 131676 Verification 10
The lower anemometers and wind vanes are mounted on booms that are designed according to the IEC
standard 61400-12-1 (edition 1) such that flow distortion is minimised [1].
The mast and some of the instruments (WF1) were installed on MM01 on 18/07/2013. In January and
February 2015, this met mast was equipped with a parallel measurement system with additional
instruments. On 24/02/2015, the refurbished MM01 was commissioned. All new anemometers were
calibrated at the wind tunnel facility of Deutsche WindGuard, Germany in November 2014. All
calibration certificates can be found in Appendix A. The uncertainty in wind measurements is minimised
through the selection of high-quality, calibrated instruments.
For the current test, the wind direction recordings from both the wind vane at 118m and 40m were
used and assigned to the closest measurement height of the cup-anemometers. Other instruments
include a rain sensor, temperature sensor, humidity sensor and a barometer.
2.3 SoDAR Location
The SODAR was installed by personnel from Vaisala Inc. at a test position at a distance of approximately
142 m to the WNW of the TSL-MM01 met mast, as specified in Table 6 and shown in Figure 3. Details
of the installation are presented in a separate report by Ecofys WTTS. The SoDAR was configured to
record wind speeds at 30, 40, 50, 60, 80, 100, 120, 140, 160, 180, and 200 m (above ground level),
which match the measurement heights of the met mast, as well as a number of other heights (not
included in this verification analysis).
The mean horizontal wind speed, mean wind direction were provided for each 10 min period at each
height. Moreover, the mean vertical wind speed and a quality column within a 10 min time interval for
each height were also recorded and provided.
Table 5: Details of the SODAR during verification campaign
Parameters Triton SoDAR 604
Test ID Triton SoDAR I
System serial number 604
Measurement range 40-200 m above ground level
Beam angles 11.5°
Timestamp interval 10 min
Timestamp UTC
Data period 05/03/2015 to 26/05/2015
Data columns recorded
Mean horizontal wind speed, Mean wind
direction, Mean vertical wind speed, TI Quality,
Quality
WIEWT15389 – Triton SoDAR 131676 Verification 11
Table 6: Coordinates of Triton SoDAR 604 measurement position
SODAR ID Longitude
[WGS84]
Latitude
[WGS84]
Easting
[RD]
Northing
[RD]
Triton 604 52.527061 E 5.584502 N 168,390 504,397
Figure 3: Triton SoDAR 604 deployed at Ecofys WTTS test site with view of TSL-MM01 142 m away.
2.4 SoDAR Orientation and Time Synchronization
The Triton SoDAR 604 was oriented towards the North (shown in Figure 4), so no offset was applied in
post-processing. It is synchronised to the Coordinated Universal Time (UTC) every day via GPS clock
and the timestamps correspond to the end of the ten-minute average. In the met mast data acquisition
system, the system clock is synchronised with a NTP time server 5 minutes before every hour. In this
way, the Triton SoDAR’s time series and the mast data’s time series can be examined and adjusted to
line up correctly and ensure that time offset between the SoDAR and met mast is always within 6
seconds, in agreement with IEC standards.
WIEWT15389 – Triton SoDAR 131676 Verification 12
Figure 4: Looking North over the Triton SoDAR I during its deployment. It was installed with 0° offset to the
magnetic north.
2.5 Valid wind direction sectors
Ecofys WTTS performed a Site Assessment investigation and internal report for the Triton’s SoDAR
position in reference with the met mast TSL-MM01 at Test Site Lelystad, in accordance with the
requirements of IEC 61400-12-1 regarding power performance measurements [1]. This analysis was
done before the start of the measurements to ensure that the location could provide adequate
conditions for the test: provision of a sufficiently large sector with predominant winds, and away from
obstacles that could interfere with the SoDAR measurement techniques.
The Obstacle Assessment has shown that some wind direction sectors must be excluded due to the
influence of neighbouring wind turbines. Three TSL wind turbines are operational to the southeast of
the met mast (WTG 1, 3 & 4), so the south-eastern sector is excluded. A row of wind turbines to the
WIEWT15389 – Triton SoDAR 131676 Verification 13
north (partly TSL, partly commercial) results in a large excluded sector as well. The excluded sectors
are shown in the figure below.
Figure 5: Excluded Sectors for TSL-MM01 and co-located SoDAR position, due to surrounding wind turbines (no other
significant obstacles); each colour represents a different wind turbine
Based on this Obstacle Assessment, the primary remaining valid sector for IEC-compliant wind
measurements is from the south-west and a smaller sector to the west, as shown in Figure 6.
Table 7: Excluded wind sectors for Triton SoDAR 604 Verification at test position
Obstacles at TSL Sector minimum [°] Sector maximum [°] Width of excluded sector [°]
Neighbouring wind turbines
0 71 71
106 189 83
286.5 360 73.5
Forests, building or power pylons
0 0 0
Mast shadowing of anemometry
The top anemometer (120 m) will not be affected by mast shadow, but is influenced by the lightning
rod. By comparison of wind speed measurements at lower heights, it was determined that the lightning
rod affected wind sectors from 140-180°. The boom-mounted anemometers will experience mast
shadowing for southerly winds. A comparison of wind speed ratios found influenced wind sectors from
160-190° (and 330-10° for V3).
Wakes of reference met mast on the measurements of the SoDAR
An additional data filter is applied to exclude more sectors influenced by the wakes of the met mast or
the neighbouring wind turbines. The potential influence of the wakes is investigated by plotting the
ratios of the met mast and the SoDAR measurements as a function of wind direction, as proposed in
[2]. Wind direction sectors with a median ratio of >1.05 or < 0.95 are excluded. In this case, no sectors
were excluded due to this filter.
191725334149576573818997105113
121129137145153161169177185193201209217225233241
249257265273281289297305313321329337345353361
369377385393401409417425433441449457465473481489497505513521529537545553561569577585593601609617625633641649657665673681689697705713
WIEWT15389 – Triton SoDAR 131676 Verification 14
Final valid wind sectors
The remaining valid sectors for the verification of the SoDAR against met mast TSL-MM01 are shown
in the figure and table below.
Figure 6: Valid Sectors for SoDAR position in reference to TSL-MM01 at Test Site Lelystad
The selection of this SoDAR position enabled the use of the sectors showed above, while keeping the
distance to the closest nearby turbines more equal and more than 330 m away. The distance to Mast
was calculated to be of 142.5 meters, which supported the informal practical criteria “SoDAR distance
= height of the mast + 20 m”.
Table 8: Valid Wind Direction Sectors for SoDAR Verification Measurement for TSL-MM01
Sector
minimum [°]
Sector
maximum [°]
Width of Sector [°]
E 71 106 35
S-NW 189 286.5 97.5
WIEWT15389 – Triton SoDAR 131676 Verification 15
3 SoDAR Verification Procedure
The SoDAR data will be compared to the met mast data on the basis of 10 minute averages.
To increase the accuracy and repeatability of the verification test, the datasets are filtered according
to the criteria described below. The filtering is carried out in Windographer using flag rules to disable
data.
The filtered data forms the basis for the data analysis, based on the IEC 61400-12-1 verification
procedure and uncertainty evaluation [1] [2] and NORSEWInD SoDAR validation criteria [3]. All data
analyses techniques are described in Section 3.3, and are performed in Excel.
3.1 Met Mast Data Filtering The met mast dataset is filtered according to the following protocols:
3.1.1 System non-availability
This category covers power outages, maintenance and other external issues. During these periods, no
data is recorded, so data does not need to be flagged or disabled. This category is defined to explain
any missing data due to external issues.
3.1.2 Wind direction recordings
Wind direction measurements are not available at all heights. Therefore, Wind Vane WF1 (at 118 m) is
linked to wind speeds of 80, 100 and 120 m. Wind Vane WF2 (at 40 m) is linked to wind speeds at 40
m.
3.1.3 Icing of instruments
During sub-zero temperatures, all met mast instruments including cup-anemometer data will be
checked for icing by inter-comparison with other heights and checks for constant output during frost.
In case suspicion of icing exists, all data will be disabled and flagged as ‘Outside operational envelope’.
3.1.4 Excluded wind sectors
All excluded sectors are filtered out, according to the valid sectors identified in Section 2.5. This
category combines mast shadowing of mast instrumentation and disturbed sectors. All wind data is
disabled if either wind vane is within an excluded sector, and flagged as ‘Excluded sectors.’
3.2 SoDAR Data Filtering
The SoDAR dataset is filtered according to the following protocols:
3.2.1 System non-availability
This category covers power outages, maintenance and other external issues. During these periods, no
data is recorded, so data does not need to be flagged or disabled. This category is defined to explain
any missing data due to external issues.
WIEWT15389 – Triton SoDAR 131676 Verification 16
3.2.2 Excluded wind sectors
All wind data is disable and flagged as ‘Excluded Sectors’ for any time stamp where either met mast
wind direction is within excluded sectors.
3.2.3 Conditions outside operational envelope
A SODAR can be disturbed by strong noise or weather conditions such as heavy snow. These periods
would be marked as “outside operational envelope”. The client provided vertical wind speed criteria to
account for invalid points due to precipitation, which was assigned to this filter.
3.2.4 Wind Quality Factor - Low signal to noise conditions
The wind quality factor is embedded in the acquisition of the SoDAR measurement data and is a function
of the signal to noise ratio and the number of valid samples collected for every 10 minute interval. It
is used to filter noisy wind data and as per manufacturer’s recommendation all timestamps where the
quality factor values are less than 90% were filtered out.
3.2.5 Turbulence Intensity Quality Factor
Since the vector wind speed is inversely proportional to the turbulence intensity estimation, high TI
values are produced at low wind speeds. In the Triton SoDAR 604 a turbulence quality factor was
defined to eliminate invalid turbulence measurements from the data set in low wind speed conditions
e.g. >3.5m/s. As per manufacturer’s recommendation, this column and all TI Quality values < 90%
were filtered out.
3.2.6 Vertical wind speed filter criteria
During rain, the Triton SoDAR 604 can interpret the falling raindrops or snowflakes as a strong vertical
wind and, as a result, the measured wind speed can be incorrect. Filtering vertical wind speeds > +/-
1.5 m/s removes any data affected by precipitation.
3.2.7 Data processing issues
Data is disabled in the event the data processing software fails to remove erroneous data.
3.3 Statistical tests
A number of statistical methods are applied to the filtered datasets, based on the requirements of IEC
61400-12-1 [2].
3.3.1 Scatter plot
The measurements of the SoDAR are plotted against the measurements of the reference anemometer,
also showing the deviations between the SoDAR and reference anemometer. The wind speed deviation
is defined as the deviation between SoDAR and met mast mean wind speeds at the same height. The
mean deviation and standard deviation of deviations is also calculated.
3.3.2 Bin averaged analysis
A bin averaging procedure is used to compare the SoDAR measurements and the reference
anemometers, using wind speed bins of 0.5 m/s between 4-16 m/s. The bin-wise deviation between
the SoDAR and reference measurements is the key result. Each wind speed bin should contain at least
3 valid pairs of data to ensure a representative SoDAR verification [2].
WIEWT15389 – Triton SoDAR 131676 Verification 17
Linear regression analysis is applied in order to evaluate the relationship in terms of horizontal mean
wind speed and mean wind direction with zero and non-zero constants at each height. Also the
coefficient of determination (R²) of the linear regression is given. Below, the formulae are given for
each parameter that is analysed.
Two-parameter linear regression
The two-parameter Ordinary Least Squares regression equation is:
kxCy
Coefficient of determination
The coefficient of determination (R²) of the linear regression is calculated as:
tot
err
SS
SSR 12
2
i
iierr fySS
2
i
itot yySS
where if is the modelled value and y is the mean of the observed data.
Wind speed deviation
The mean and standard deviation of the wind speed deviation is also calculated for each wind speed
bin of 0.5 m/s.
3.3.3 Wind direction
Similarly, a bin averaging procedure is used to compare the SoDAR wind direction measurements and
the reference wind vanes, using wind direction bins of 10° (based on the test uncertainty shown in
Section 2.2). A two-parameter Ordinary Least Squares linear regression is used, in order to identify
any offset from zero.
3.3.4 Sensitivity tests
Several sensitivity tests are run to find any possible relationship with external conditions, including
SoDAR vertical wind speed, met mast mean wind speed, turbulence intensity and precipitation.
3.3.5 Uncertainty analysis
After all sensitivity tests and statistical analyses, Ecofys analyses the uncertainty of the wind
measurements of this SoDAR. The uncertainty calculations are based on IEC 61400-12-1 (ed 2) which
is still in draft [2]. If information is absent in the draft, formulae and definitions were taken from the
first edition [1].
The uncertainty resulting from the SoDAR verification test is divided into 5 separate uncertainties, as
summarised below:
1. Reference uncertainty (in anemometry)
a. Wind tunnel calibration
WIEWT15389 – Triton SoDAR 131676 Verification 18
b. Cup anemometer effects according to the anemometer classification
c. Cup anemometer mounting effects
i. Mast shadowing
ii. Boom distortion
iii. Lightning rod distortion
d. Uncertainty of any applied mast correction
2. Mean deviation of the SoDAR measurements and the reference anemometry measurements
3. Standard deviation of the measurement of the SoDAR
4. Uncertainty in mounting effects during the verification test
5. Uncertainty of the SoDAR due to non-homogeneous flow within the measurement volume,
during the verification test
The uncertainties are combined as the square root of the sum of squares:
𝑈𝑡𝑜𝑡𝑎𝑙 = √𝑈12 + 𝑈22 + 𝑈32 + 𝑈42 + 𝑈52
The calculated uncertainty refers to the uncertainty in the performance verification test. The total
uncertainty for future wind measurement campaigns will also include components relating to site-
specific mounting and flow condition on-site, and uncertainty resulting from the classification of the
SoDAR (sensitivity to environmental variables). These uncertainties should be assessed as part of the
wind resource assessments.
3.3.6 NORSEWInD criteria
Some additional statistical tests are performed to compare with the NORSEWInD criteria [3].
The linear regression, including coefficient of determination, is re-calculated for three different wind
speed ranges without any bin-averaging. First, all wind data is used, then two smaller ranges are
filtered for 4-8 m/s and 8-12 m/s. The slopes of the linear regression fit are compared for the two
smaller ranges, in order to identify non-linearity in the system performance.
Also, the wind speed deviation is assessed in terms of absolute error.
Single-parameter linear regression
The NORSEWInD criteria are based on a single parameter Ordinary Least Squares regression model
according to the following equation:
mxy
y = SoDAR measurement
x = met mast measurement
WIEWT15389 – Triton SoDAR 131676 Verification 19
4 Results & Discussion
The results for the Triton SoDAR 604 verification are presented below. The data coverage is shown,
followed by linear regression of the wind speed and direction measurements. A sensitivity analysis is
shown against several environmental factors. Finally, the verification uncertainty analysis and
validation against NORSEWInD criteria are presented.
4.1 Data coverage
The data filtering (described in the previous chapter) reduced the number of data points by about half,
as seen in Table 9. The disabled data was primarily due to wind directions from excluded sectors, as
well as some SoDAR data that was flagged as per filters described above in Chapter 3.
Table 9 : Number of data points and data coverage before and after filtering for the Triton SoDAR 604.
Height [m]
Number of data points
Before filtering After filtering
SoDAR Met mast SoDAR Met mast
40 12,048 12,241 5,130 6,244
80 12,048 12,241 5,477 6,335
100 12,048 12,241 5,512 6,335
120 12,048 12,241 5,508 6,335
WIEWT15389 – Triton SoDAR 131676 Verification 20
Time series of the valid data, in terms of mean wind speed, wind speed deviation and wind
direction are plotted in the graphs below.
Figure 7. Time series of the Triton SoDAR 604 and met mast mean wind speed at 120m, 100, 80 and 40m.
Figure 8. Time series of the Triton SoDAR 604 and met mast wind speed deviation at 120m, 100, 80 and 40m.
Figure 9. Time series of the Triton SoDAR 604 and met mast mean wind direction at 120m, 100, 80 and 40m.
0
5
10
15
20
25
30
21/02/2015 07/03/2015 21/03/2015 04/04/2015 18/04/2015 02/05/2015 16/05/2015 30/05/2015 13/06/2015
Mean
win
d s
peed
[m
/s]
Vmean SoDAR 120m
Vmean met mast 120m
Vmean SoDAR 100m
Vmean met mast 100m
Vmean SoDAR 80m
Vmean met mast 80m
Vmean SoDAR 40m
Vmean met mast 40m
-4
-3
-2
-1
0
1
2
21/02/2015 07/03/2015 21/03/2015 04/04/2015 18/04/2015 02/05/2015 16/05/2015 30/05/2015 13/06/2015
Win
d s
peed
devia
tio
n [
m/
s]
Wind speed deviation 120 m
Wind speed deviation 100 m
Wind speed deviation 80 m
Wind speed deviation 40 m
0
90
180
270
360
21/02/201507/03/201521/03/201504/04/201518/04/201502/05/201516/05/201530/05/201513/06/2015
Win
d d
irecti
on
Wind direction SoDAR 120 m
Wind direction met mast 118 m
wind direction SoDAR 100 m
Wind direction SoDAR 80 m
wind direction met mast 40 m
Wind direction SoDAR 40 m
WIEWT15389 – Triton SoDAR 131676 Verification 21
4.2 Wind speed verification
The mean wind speed measured by the Triton SoDAR is compared to the concurrent met mast
measurements in a scatter plot (see Figure 10, Figure 11, Figure 12 and Figure 13) and a single
parameter “Ordinary Least Squares” (OLS) linear regression is applied. The slope of the linear
regression and the corresponding coefficients of determination (R²) are shown. The plots show a very
good correlation between the SoDAR and met mast wind speed measurements.
The wind speed deviation is defined as the SoDAR mean wind speed minus the met mast mean wind
speed. This is also plotted (in m/s) in the scatter plots below. The statistical distribution of the wind
speed deviations has been derived, as shown in Table 10.
Figure 10: Linear regression of mean wind speed at 120m for the Triton SoDAR 604.
WIEWT15389 – Triton SoDAR 131676 Verification 22
Figure 11: Linear regression of mean wind speed at 100m for the Triton SoDAR 604.
Figure 12: Linear regression of mean wind speed at 80m for the Triton SoDAR 604.
WIEWT15389 – Triton SoDAR 131676 Verification 23
Figure 13: Linear regression of mean wind speed at 40m for the Triton SoDAR 604.
Table 10: Statistical parameters of wind speed deviation for each reference height for Triton SoDAR 604.
Height [m]
Mean of Wind Speed
Deviations
[m/s]
Standard Deviation of
Wind Speed Deviations
[m/s]
40 0.025 0.525
80 0.023 0.476
100 -0.020 0.434
120 -0.020 0.452
4.3 Wind direction verification
Similarly, a linear fit is applied to scatter plots of the wind direction, as shown in Figure 14 through
Figure 17. A two-parameter Ordinary Least Squares linear regression is used, in order to identify any
offset from zero, with slopes, offsets and coefficients of determination (R²) plotted in the graphs.
The correlation between SoDAR and met mast binned wind directions is excellent, with a slope very
close to unity and an offset of approximately +3° which is within the mounting uncertainty of the
SoDAR and wind vanes. Thus, there is no apparent offset and hence none has been corrected for.
WIEWT15389 – Triton SoDAR 131676 Verification 24
Figure 14: Linear regression of mean wind direction at 120m for the Triton SoDAR 604.
Figure 15: Linear regression of mean wind direction at 100m for the Triton SoDAR 604.
WIEWT15389 – Triton SoDAR 131676 Verification 25
Figure 16: Linear regression of mean wind direction at 80m for the Triton SoDAR 604.
Figure 17: Linear regression of mean wind direction at 40m for the Triton SoDAR 604.
WIEWT15389 – Triton SoDAR 131676 Verification 26
4.4 Sensitivity tests
Several sensitivity tests are run to find any possible relationship with external conditions. The wind
speed deviation at 120 m is plotted against SoDAR vertical wind speed, met mast mean wind speed,
turbulence intensity and precipitation, as seen in Figure 18. The plots show negligible correlation,
indicating that filtered SoDAR data is insensitive to vertical wind speed, mean wind speed, turbulence
intensity or rain. Most periods with any precipitation are efficiently identified and filtered out (as well
as larger vertical wind speeds) with the SoDAR filters.
a) SoDAR vertical mean wind speed
b) Reference wind speed
c) Turbulence Intensity
d) Wind speed deviation against precipitation
Figure 18: Sensitivity tests of wind speed deviation at 120m plotted against other factors, for the Triton SoDAR 604.
WIEWT15389 – Triton SoDAR 131676 Verification 27
4.5 Verification uncertainty analysis
For each height an uncertainty analyses was applied to each wind speed bin of 0.5 m/s between 4 and
16 m/s. For the all bins, a sufficient number of valid data points were collected. A detailed overview
table can be found in Appendix C.
This analysis shows low uncertainty in the SoDAR wind speed measurements, in line with first class
anemometry. The uncertainty is increased at higher wind speed bins, partly due to relatively fewer
data points. The uncertainty levels are higher for the 40 meters measurement height, due to the
formula for the uncertainty due to variation in flow across the site, which is a function of the
measurement height and the distance between the SoDAR and mast (see Appendix B).
The calculated uncertainty can be used directly in wind resource assessments with this device, together
with the classification uncertainty and site-specific uncertainty components.
Figure 19: Uncertainty analysis of SoDAR wind speed measurements at 120 m for the Triton SoDAR 604.
WIEWT15389 – Triton SoDAR 131676 Verification 28
Figure 20: Uncertainty analysis of SoDAR wind speed measurements at 100 m for the Triton SoDAR 604.
Figure 21: Uncertainty analysis of SoDAR wind speed measurements at 80m m for the Triton SoDAR 604.
WIEWT15389 – Triton SoDAR 131676 Verification 29
Figure 22: Uncertainty analysis of SoDAR wind speed measurements at 40 m for the Triton SoDAR 604.
WIEWT15389 – Triton SoDAR 131676 Verification 30
4.6 NORSEWInD criteria validation
The results of the verification campaign are also validated against the NORSEWInD criteria outlined in
Section 1.3. Table 11 shows that this device meets most of the NORSEWInD criteria and all of the
Ecofys WTTS acceptance thresholds for field measurements, which are highlighted in bold, with passing
results highlighted in green, failing results in red and marginal passes shown in yellow. A marginal pass
is a result that deviates from the criteria by a small amount.
The overall high correlations and very good linear regression fit for the entire data set indicate that the
Triton SoDAR 604 is functioning properly, with high accuracy and there are no underlying issues
present. The validation of the Triton SoDAR 604 shows that this device is suitable for field
measurements based on the Ecofys WTTS acceptance thresholds.
Table 11: Criteria analysis for the Triton SoDAR 604.
Citerium Category
NORSEWInD Acceptance Threshold for Replacement
of IEC-compliant Cup Anemometry
Ecofys WTTS Acceptance Threshold
for Field Measurements
120m 100m 80m 40m
Number of valid data points
4-8m/s >200 2,630 2,898 3,168 3,326
8-12 m/s >200 2,064 1,869 1,620 1,134
ALL >600 5,508 5,512 5,477 5,130
Percentage of data points that exceed 0.5 m/s absolute error
<10% n/a 18.8% 17.5% 18.9% 21.9%
Linear regression slope
4-8m/s 0.98-1.01 n/a 1.003 0.999 1.004 1.004
8-12 m/s 0.98-1.01 n/a 0.999 0.999 1.011 1.017
Variation in slope
<0.015 n/a 0.003 0.000 0.008 0.012
ALL 0.98-1.01 0.98-1.02 0.995 0.996 1.000 1.000
Linear regression – R2
4-8m/s >0.98 n/a 0.914 0.918 0.911 0.893
8-12 m/s >0.98 n/a 0.866 0.867 0.841 0.746
ALL >0.98 >0.95 0.975 0.976 0.972 0.965
WIEWT15389 – Triton SoDAR 131676 Verification 31
5 Conclusions
On request of Vaisala Inc., Ecofys Wind Turbine Testing Services (WTTS) carried out a verification of a
Triton SoDAR 604 (Triton with logger number 604) against met mast TSL-MM01 at the WTTS Test Site
Lelystad (TSL).
The 120 m TSL-MM01 is fully IEC and MEASNET compliant and equipped with cup-anemometers at 40,
80, 100 and 120 m. In addition, wind vanes for wind direction measurements are located at 87 and
118 m. The SoDAR was installed 142.5 meters WNW from MM01 and collected wind data at four
common heights with the met mast. The SoDAR verification campaign lasted from 05/03/2015 to
26/05/2015.
All measured data was collected at the specified location and filtered to ensure entirely valid datasets.
The verification analysis is based on the IEC standard 61400-12-1 (ed 2, draft [2]). The Triton SoDAR
604 verification test covers a full three months of collocated measurements during spring and summer
2015, and has permitted the collection of over 5,000 concurrent valid data points. This resulting dataset
is significantly larger than the minimum required, providing a robust basis for uncertainty calculations
in each of the 0.5m/s wind speed bins up to 16 m/s and demonstrates the SoDAR’s performance across
seasons.
Linear regression between the overall Triton SoDAR 604 and met mast recordings showed consistent,
highly-correlated measurements with slopes near unity. The calculated uncertainty in the SoDAR wind
speed measurements is relatively low, in line with high quality anemometry for the majority of the wind
speed bins. The calculated uncertainty tables (in Appendix C) can be used directly in wind resource
assessments, together with the classification uncertainty and site-specific uncertainty components.
Sensitivity tests of the wind speed deviation revealed that the wind speed deviation shows no significant
linear correlation to external conditions: vertical wind speed, horizontal wind speed, turbulence
intensity or rain. This indicates that the filtering techniques recommended by the manufacturer are
precise, and make the Triton SoDAR 604 relatively insensitive to these factors.
The SODAR measurements were also validated against NORSEWiND criteria [3] and meets the Ecofys
WTTS acceptance thresholds for field measurements. There is scatter in the individual recordings, which
may be partly due to the distance between the SoDAR and mast during the test. The overall high
correlations and very good linear regression fit indicate that the SoDAR is functioning properly with
high accuracy.
As a result of this test, Ecofys WTTS judges that the SoDAR Triton 604 is suitable for field
measurements. The SODAR will measure the long-term mean wind speeds with accuracy comparable
to cup anemometry in flat terrain. However, when analysing Triton data sets, due consideration should
be made for the standard deviation of 10-minute values characteristic to the system.
WIEWT15389 – Triton SoDAR 131676 Verification 32
References
[1] IEC 61400-12-1 (ed 1.0), ‘Wind turbines – Part 12-1: Power performance measurements of electricity producing wind turbines’, 2005
[2] IEC, IEC 61400-12-1 Ed. 2 (642/DCV) ‘Committee Draft - Power performance measurements of electricity producing wind turbines’, 2013
[3] Kindler, D., Courtney, M., Oldroyd, A., 2009, 'Testing and calibration of various SoDAR remote
sensing devices for a 2 year offshore wind measurement campaign', EWEC 2009.
[4] Lindelöw-Marsden, P. ‘Upwind D1: Uncertainties in wind assessment with SODAR’. Risø DTU, 2009, Risø-R-1681(EN)
[5] Topographische Dienst, ‘Compact Provincie Atlas 1:50,000 – Utrecht/Flevoland’, 1997
[6] Deutsche Windguard, 2008, ‘Summary of cup-anemometer classification according to IEC 61400-12-1 (2005-12) Classification scheme’, AK081662.01S
WIEWT15389 – Triton SoDAR 131676 Verification 33
Appendix A Anemometer calibration certificates Top cup-anemometer A1 at 120 m
WIEWT15389 – Triton SoDAR 131676 Verification 34
Cup-anemometer A2 at 100 m
WIEWT15389 – Triton SoDAR 131676 Verification 35
Cup-anemometer A3 at 80 m
WIEWT15389 – Triton SoDAR 131676 Verification 36
Cup-anemometer A4 at 40 m
WIEWT15389 – Triton SoDAR 131676 Verification 37
Appendix B Uncertainty analysis
As described in Annex L of IEC 61400-12-1 (edition 2 draft) [2], the uncertainty in the SoDAR wind
measurements can be divided into 6 separate uncertainties, as summarised below:
1. Reference uncertainty (in anemometry)
a. Wind tunnel calibration
b. Cup anemometer effects according to the anemometer classification
c. Cup anemometer mounting effects
i. Mast shadowing
ii. Boom distortion
iii. Lightning rod distortion
d. Uncertainty of any applied mast correction
2. Mean deviation of the SoDAR measurements and the reference anemometry measurements
3. Standard deviation of the measurement of the SoDAR
4. Uncertainty in mounting effects at the verification test
5. Uncertainty of the SoDAR due to non-homogeneous flow within the measurement volume,
during the verification test
6. Uncertainty due to variation in flow across the site
The uncertainties are combined as the square root of the sum of squares:
𝑈𝑡𝑜𝑡𝑎𝑙 = √𝑈12 + 𝑈22 + 𝑈32 + 𝑈42 + 𝑈52
1. Reference uncertainty (Met mast anemometer)
The following uncertainty components are considered for the evaluation of the reference sensor
uncertainty.
1. a) Wind tunnel calibration
The uncertainty in wind tunnel calibration is defined as the root sum square of standard error in linear
fit in the calibration certificates of the met mast and the wind tunnel accuracy. The calibration
certificates for the anemometers at met mast TSL-MM01 can be found in Appendix A.
1. b) Cup anemometer effects according to the anemometer classification
This uncertainty relates to the sensitivity of the cup anemometer to ranges of environmental
parameters per terrain class. Valid ranges of environmental variables are defined for Class A and B in
Annex I of [1], as shown in Table 12. Test Site Lelystad meets all of the criteria of a Class A site, unless
temperatures drop below 0 or turbulence intensity is incidentally exceeding the class A thresholds.
WIEWT15389 – Triton SoDAR 131676 Verification 38
Table 12: Influence parameter ranges (based on 10 min average) of Classes A and B [1]
The sensitivity of several cup anemometer models was calculated in the Deutsche Windguard study
[6]. For a Thies First Class cup anemometer (as used at TSL-MM01), in Class A conditions, the class
number k equals to 0.9, and in Class B conditions (such as periods with temperatures below zero), k is
3.0. This class number k is inserted into the following equation to derive the standard uncertainty due
to cup-anemometer type and environmental conditions during the verification test:
𝑢𝑉2,𝑖 = (0,05 𝑚 𝑠⁄ + 0,005 ∙ 𝑈𝑖) ∙ 𝑘/√3
c) Cup anemometer mounting effects
The cup anemometers are influenced by flow distortion of the met mast’s components:
For the top anemometer: lightning rod distortion ((and interference with each other)
For the cup anemometers below top: mast distortion and boom distortion
Mast distortion depends upon the type of mast and its solidity, the drag of the individual members, the
orientation of the wind and the separation of the measurement point from the mast. Therefore in Annex
G.4 of [2], mast distortion functions are defined for different mast types. Based on met mast design
drawings, met mast TSL-MM01 has a solidity of 0.15 and an R/L (roughly boom length over leg
distance) of 5.2 which results in a Ct of 0.27. Entering these values into the equation below results in
a standard uncertainty of about 0.3% of the measured 10 min wind speed at every time interval.
𝑈𝑑 = 1 − (0.062𝐶𝑡2 + 0.076𝐶𝑡) ∙ (
𝐿
𝑅− 0.082)
The flow distortions due to the booms equals 0.5% as this IEC-compliant met mast was designed such
that its boom distortion is kept below 0.5%.
Lightning rod distortion can be ignored if the following conditions are met [2]:
1. The top cup-anemometer is separated horizontally at least 30 times the lightning rod diameter
from the lightning rod. As separation equals 30.8 times for TSL-MM01, this criteria is met
2. The cup-anemometer is not in the wake of the lightning rod. As the sectors disturbed by the
lightning rod overlap with the excluded sectors, no wind data disturbed by the lightning rod is
in the data set anymore.
WIEWT15389 – Triton SoDAR 131676 Verification 39
d) Uncertainty of any applied mast correction
This uncertainty defines any uncertainties in applied mast corrections. Since no correction factors were
used for the wind speed data, this uncertainty is excluded.
2. Mean deviation of the SoDAR measurements and the reference anemometry
measurements
This refers to the mean deviation of the remote sensing device measurements and the reference sensor
measurements for each bin of 0.5 m/s as defined in Annex L4.2.1 of [2]. The standard uncertainty is
derived from the difference between remote sensing device mean wind speed and met mast mean wind
speed. To obtain a percentage, the standard uncertainty is divided by the met mast mean wind speed.
Therefore the mean deviation uncertainty tends to be higher at lower wind speeds.
3. Precision of the measurement of the SODAR
The precision of the measurement of the remote sensing device calculated as the standard deviation
of the measurements divided by the square root of the number of data per bin as defined in Annex
L4.2.1 of [2].
4. Uncertainty in mounting effects at the verification test
This uncertainty is related to non-ideal levelling of the device and shall be estimated by us according
to Annex L4.5 in [2]. This is assumed to be 0% based on stable foundations for the Ecofys WTTS test
pad.
5. Uncertainty of the SoDAR due to non-homogeneous flow within the measurement volume,
during the verification test
Considering the homogenous terrain conditions within the valid sectors, this uncertainty component is
considered negligible and therefore set to 0%.
6. Uncertainty due to variation in flow across the site
There is an uncertainty due to variation of flow conditions between the centre of the remote sensing
device and the met mast. According to L4.2.1 in [2], this can be calculated as 1% times the separation
distance divided by the measurement height.
WIEWT15389 – Triton SoDAR 131676 Verification 40
Table 13: Uncertainties for TSL-MM01 used for this SoDAR verification
Uncertainty
category
Uncertainty
subcategory Parameter Value
Reference uncertainty
(in anemometry)
See uncertainty table, defined per wind speed bin of
0.5 m/s
Wind tunnel
calibration
Calibration
standard error
in linear fit
and the wind
tunnel
accuracy for
k=1
Defined per wind speed bin of 0.5%
0.014 m/s (at 120 m) and 0.025 m/s
0.015 m/s (at 100 m) and 0.025 m/s
0.016 m/s (at 80 m) and 0.025 m/s
0.021 m/s (at 40 m) and 0.025 m/s
Cup anemometer
effects according
to the anemometer
classification
k 0.9
Cup anemometer
mounting effects Uncertainty
0% for top anemometer (data possibly affected by
lightning rod distortion are disabled)
0.5% for boom distortion
Uncertainty of any
applied mast
correction
0.0%
Mean deviation of the SoDAR
measurements and the reference
anemometry measurements
See uncertainty table, defined per wind speed bin of
0.5 m/s
Standard deviation of the
measurement of the SoDAR
See uncertainty table, defined per wind speed bin of
0.5 m/s
Uncertainty in mounting effects
at the verification test Uncertainty 0.0%
Uncertainty of the SoDAR due to
non-homogeneous flow within
the measurement volume,
during the verification test
Uncertainty 0.0%
Uncertainty due to variation in
flow across the site Uncertainty
Defined per height, based on the separation distance
between met mast and remote sensing device:
1.19%s (at 120 m)
1.42% (at 100 m)
1.78% (at 80 m)
3.56% (at 40 m)
WIEWT15389 – Triton SoDAR 131676 Verification 41
Appendix C Detailed uncertainty analysis
Data coverage
During this verification campaign, the met mast was fully available for measurements. The SoDAR was
operating 100% of the time.
Table 14: Logbook of technical issues and external problems of the SoDAR and met mast.
Date and time Duration Issue Details
n/a
The breakdown of the data points flagged by each filtering criteria is shown in Table 15. The same data
point can be flagged for different criteria. Therefore the sum of all criteria exceeds the total number of
data points possible within the measurement period. All flagged data was disabled and excluded from
the verification.
The conditions specific to this verification campaign affect the count of several categories and are
therefore not necessarily representative of the long-term. System non-availability for instance is
affected by external power outages, while low signal to noise conditions are raised due to rain and
strong wakes from wind turbines in disturbed sectors.
Table 15: Number of data points flagged for each filtering criteria for each height. Numbers include double-counts of
data points, as one data point can be flagged for several categories. Environmental conditions at the test site affect
the count of system non-availability (e.g. due to power outages) and low signal to noise conditions (strong wakes
from several nearby wind turbines), which are not representative of a random measurement campaign.
Height [m]
System non-availability
Excluded wind sectors
Low SNR, SoDAR Filters & Met Mast Icing
Data processing issues
Conditions outside
operational envelope & wind speed
intercomparison
40 192 0 5951 5995 2340 0 0 0 0 0
80 192 0 5854 5903 1953 0 0 0 0 0
100 192 0 5854 5903 1897 0 0 0 0 0
120 192 0 5854 5903 1875 0 0 0 0 0
After filtering, all wind speed bins should have at least the required amount of wind data (3 valid data
points per bin). Sometimes certain bins do not meet these criteria. If the two adjacent bins had 3 or
more data points, the overall uncertainty could be interpolated. Otherwise no uncertainty is shown for
the entire wind speed bin.
Uncertainty Analysis
The detailed results of the uncertainty calculations, based on the IEC 61400-12-1 methods, are
presented below. The uncertainty figure in the tables below are based on a coverage factor of 1 which
implies that the level of confidence is 68%.
WIEWT15389 – Triton SoDAR 131676 Verification 42
Table 16 - Uncertainty analysis at 120 m for the Triton SoDAR 604
Bin
Vcu
p
VR
SD
Nu
mb
er o
f
data
sets
VR
SD m
ax
VR
SD m
in
VR
SD-s
td
VR
SD-s
td/√n
Mean
devia
tio
n
Vcu
p
un
certa
inty
Mou
nti
ng
un
certa
inty
Flo
w
un
certa
inty
VR
SD
un
certa
inty
4 4.010 3.975 169 7.60 3.42 0.386 0.030 -0.9% 1.2% 0.0% 1.2% 2.0%
4.5 4.525 4.486 222 6.06 3.31 0.346 0.023 -0.9% 1.0% 0.0% 1.2% 1.9%
5 5.012 4.963 278 6.23 3.77 0.344 0.021 -1.0% 1.0% 0.0% 1.2% 1.9%
5.5 5.501 5.530 312 7.13 4.52 0.392 0.022 0.5% 0.9% 0.0% 1.2% 1.6%
6 5.997 6.053 311 7.81 4.70 0.398 0.023 0.9% 0.8% 0.0% 1.2% 1.8%
6.5 6.496 6.508 324 8.56 5.00 0.376 0.021 0.2% 0.8% 0.0% 1.2% 1.5%
7 6.992 7.014 435 8.19 5.62 0.362 0.017 0.3% 0.8% 0.0% 1.2% 1.5%
7.5 7.496 7.537 440 9.52 6.22 0.406 0.019 0.5% 0.7% 0.0% 1.2% 1.5%
8 7.994 8.005 423 9.74 6.90 0.376 0.018 0.1% 0.7% 0.0% 1.2% 1.4%
8.5 8.494 8.514 352 9.77 6.78 0.379 0.020 0.2% 0.7% 0.0% 1.2% 1.4%
9 8.989 8.961 335 10.79 7.67 0.422 0.023 -0.3% 0.6% 0.0% 1.2% 1.4%
9.5 9.493 9.478 247 11.84 8.28 0.441 0.028 -0.2% 0.6% 0.0% 1.2% 1.4%
10 10.014 9.978 249 11.36 8.11 0.482 0.031 -0.4% 0.6% 0.0% 1.2% 1.4%
10.5 10.482 10.411 240 12.05 7.82 0.511 0.033 -0.7% 0.6% 0.0% 1.2% 1.5%
11 10.985 11.006 214 12.49 9.56 0.527 0.036 0.2% 0.6% 0.0% 1.2% 1.4%
11.5 11.483 11.566 152 13.40 10.54 0.555 0.045 0.7% 0.5% 0.0% 1.2% 1.5%
12 11.990 11.976 141 13.88 10.78 0.584 0.049 -0.1% 0.5% 0.0% 1.2% 1.4%
12.5 12.498 12.513 127 13.96 11.24 0.572 0.051 0.1% 0.5% 0.0% 1.2% 1.4%
13 12.981 12.866 99 14.59 11.79 0.552 0.055 -0.9% 0.5% 0.0% 1.2% 1.6%
13.5 13.507 13.255 61 14.88 12.30 0.549 0.070 -1.9% 0.5% 0.0% 1.2% 2.3%
14 13.986 13.765 75 15.60 10.72 0.804 0.093 -1.6% 0.5% 0.0% 1.2% 2.1%
14.5 14.528 14.372 51 15.97 12.10 0.689 0.096 -1.1% 0.5% 0.0% 1.2% 1.8%
15 14.982 14.750 45 16.27 13.38 0.598 0.089 -1.6% 0.5% 0.0% 1.2% 2.1%
15.5 15.500 15.243 44 16.50 13.84 0.614 0.093 -1.7% 0.5% 0.0% 1.2% 2.2%
16 16.047 15.613 17 17.12 14.42 0.736 0.179 -2.7% 0.5% 0.0% 1.2% 3.2%
WIEWT15389 – Triton SoDAR 131676 Verification 43
Table 17 - Uncertainty analysis at 100 m for the Triton SoDAR 604.
Bin
Vcu
p
VR
SD
Nu
mb
er o
f
data
sets
VR
SD m
ax
VR
SD m
in
VR
SD-s
td
VR
SD-s
td/√n
Mean
devia
tio
n
Vcu
p
un
certa
inty
Mou
nti
ng
un
certa
inty
Flo
w
un
certa
inty
VR
SD
un
certa
inty
4 4.006 3.968 178 4.69 3.32 0.279 0.021 -0.9% 1.3% 0.0% 1.4% 2.2%
4.5 4.497 4.460 263 6.19 3.41 0.331 0.020 -0.8% 1.2% 0.0% 1.4% 2.1%
5 4.995 4.955 283 6.04 3.79 0.330 0.020 -0.8% 1.1% 0.0% 1.4% 2.0%
5.5 5.508 5.506 341 8.85 4.65 0.364 0.020 0.0% 1.0% 0.0% 1.4% 1.8%
6 6.004 5.988 359 7.36 4.82 0.328 0.017 -0.3% 1.0% 0.0% 1.4% 1.8%
6.5 6.506 6.501 425 8.81 5.60 0.363 0.018 -0.1% 0.9% 0.0% 1.4% 1.7%
7 6.997 7.026 485 9.64 5.46 0.396 0.018 0.4% 0.9% 0.0% 1.4% 1.8%
7.5 7.482 7.463 443 8.97 6.16 0.376 0.018 -0.3% 0.9% 0.0% 1.4% 1.7%
8 7.993 8.019 391 10.07 6.88 0.399 0.020 0.3% 0.9% 0.0% 1.4% 1.7%
8.5 8.502 8.489 325 9.74 7.58 0.372 0.021 -0.2% 0.8% 0.0% 1.4% 1.7%
9 8.990 8.906 287 10.14 7.43 0.420 0.025 -0.9% 0.8% 0.0% 1.4% 1.9%
9.5 9.485 9.462 253 11.33 7.87 0.439 0.028 -0.2% 0.8% 0.0% 1.4% 1.7%
10 9.995 9.993 236 11.46 8.57 0.476 0.031 0.0% 0.8% 0.0% 1.4% 1.7%
10.5 10.502 10.508 204 12.52 8.23 0.579 0.041 0.1% 0.8% 0.0% 1.4% 1.7%
11 10.986 11.061 168 12.58 9.84 0.508 0.039 0.7% 0.8% 0.0% 1.4% 1.8%
11.5 11.506 11.497 152 13.41 10.28 0.527 0.043 -0.1% 0.7% 0.0% 1.4% 1.6%
12 11.994 12.077 121 13.81 10.83 0.616 0.056 0.7% 0.7% 0.0% 1.4% 1.8%
12.5 12.499 12.474 100 14.01 11.35 0.549 0.055 -0.2% 0.7% 0.0% 1.4% 1.7%
13 12.981 13.024 52 14.25 11.90 0.547 0.076 0.3% 0.7% 0.0% 1.4% 1.7%
13.5 13.487 13.342 76 15.43 9.85 0.766 0.088 -1.1% 0.7% 0.0% 1.4% 2.0%
14 14.005 13.911 61 16.12 12.35 0.708 0.091 -0.7% 0.7% 0.0% 1.4% 1.8%
14.5 14.474 14.361 46 15.31 12.73 0.581 0.086 -0.8% 0.7% 0.0% 1.4% 1.9%
15 15.005 14.871 48 15.69 13.78 0.468 0.068 -0.9% 0.7% 0.0% 1.4% 1.9%
15.5 15.489 15.223 20 16.05 14.23 0.478 0.107 -1.7% 0.7% 0.0% 1.4% 2.4%
16 15.961 15.513 11 16.72 14.54 0.727 0.219 -2.8% 0.7% 0.0% 1.4% 3.5%
WIEWT15389 – Triton SoDAR 131676 Verification 44
Table 18 - Uncertainty analysis at 80 m for the Triton SoDAR 604.
Bin
Vcu
p
VR
SD
Nu
mb
er o
f
data
sets
VR
SD m
ax
VR
SD m
in
VR
SD-s
td
VR
SD-s
td/√n
Mean
devia
tio
n
Vcu
p
un
certa
inty
Mou
nti
ng
un
certa
inty
Flo
w
un
certa
inty
VR
SD
un
certa
inty
4 4.015 3.993 242 5.45 3.38 0.315 0.020 -0.5% 1.3% 0.0% 1.8% 2.3%
4.5 4.494 4.443 266 5.47 3.38 0.305 0.019 -1.1% 1.2% 0.0% 1.8% 2.5%
5 5.008 4.976 365 6.07 3.41 0.360 0.019 -0.6% 1.1% 0.0% 1.8% 2.2%
5.5 5.512 5.532 399 7.72 4.10 0.372 0.019 0.3% 1.0% 0.0% 1.8% 2.1%
6 6.006 6.027 439 8.13 4.98 0.362 0.017 0.4% 1.0% 0.0% 1.8% 2.1%
6.5 6.492 6.525 512 8.06 5.45 0.346 0.015 0.5% 0.9% 0.0% 1.8% 2.1%
7 6.982 7.034 478 8.81 5.75 0.404 0.018 0.8% 0.9% 0.0% 1.8% 2.2%
7.5 7.493 7.524 399 9.62 6.54 0.386 0.019 0.4% 0.9% 0.0% 1.8% 2.0%
8 7.992 8.056 340 9.63 7.07 0.438 0.024 0.8% 0.9% 0.0% 1.8% 2.2%
8.5 8.480 8.502 286 10.06 7.35 0.453 0.027 0.3% 0.8% 0.0% 1.8% 2.0%
9 8.972 9.056 275 10.62 7.85 0.475 0.029 0.9% 0.8% 0.0% 1.8% 2.2%
9.5 9.491 9.580 218 10.87 8.34 0.506 0.034 0.9% 0.8% 0.0% 1.8% 2.2%
10 10.006 10.189 212 11.86 8.88 0.551 0.038 1.8% 0.8% 0.0% 1.8% 2.7%
10.5 10.489 10.647 178 12.27 8.50 0.563 0.042 1.5% 0.8% 0.0% 1.8% 2.5%
11 10.999 11.158 144 13.17 9.77 0.599 0.050 1.4% 0.8% 0.0% 1.8% 2.5%
11.5 11.475 11.633 96 13.21 10.27 0.633 0.065 1.4% 0.7% 0.0% 1.8% 2.4%
12 11.985 12.067 90 13.63 10.66 0.571 0.060 0.7% 0.7% 0.0% 1.8% 2.1%
12.5 12.484 12.544 67 14.25 10.98 0.678 0.083 0.5% 0.7% 0.0% 1.8% 2.1%
13 13.008 12.974 68 14.89 12.13 0.596 0.072 -0.3% 0.7% 0.0% 1.8% 2.0%
13.5 13.534 13.602 57 15.90 11.87 0.781 0.104 0.5% 0.7% 0.0% 1.8% 2.1%
14 13.979 14.067 54 15.89 12.78 0.716 0.097 0.6% 0.7% 0.0% 1.8% 2.1%
14.5 14.466 14.318 34 16.29 13.19 0.595 0.102 -1.0% 0.7% 0.0% 1.8% 2.3%
15 14.938 14.849 24 16.04 13.74 0.552 0.113 -0.6% 0.7% 0.0% 1.8% 2.1%
15.5 15.524 15.253 16 16.72 14.06 0.738 0.185 -1.7% 0.7% 0.0% 1.8% 2.8%
16 15.980 15.620 6 17.78 13.79 1.425 0.582 -2.3% 0.7% 0.0% 1.8% 4.7%
WIEWT15389 – Triton SoDAR 131676 Verification 45
Table 19 - Uncertainty analysis at 40 m for the Triton SoDAR 604.
Bin
Vcu
p
VR
SD
Nu
mb
er o
f
data
sets
VR
SD m
ax
VR
SD m
in
VR
SD-s
td
VR
SD-s
td/√n
Mean
devia
tio
n
Vcu
p
un
certa
inty
Mou
nti
ng
un
certa
inty
Flo
w
un
certa
inty
VR
SD
un
certa
inty
4 4.025 3.955 351 5.26 3.37 0.303 0.016 -1.7% 1.3% 0.0% 3.6% 4.2%
4.5 4.495 4.442 501 7.25 3.42 0.384 0.017 -1.2% 1.2% 0.0% 3.6% 4.0%
5 4.999 4.961 514 6.84 3.83 0.406 0.018 -0.8% 1.1% 0.0% 3.6% 3.8%
5.5 5.500 5.500 511 7.96 4.07 0.402 0.018 0.0% 1.1% 0.0% 3.6% 3.7%
6 5.997 6.003 430 8.20 4.74 0.402 0.019 0.1% 1.0% 0.0% 3.6% 3.7%
6.5 6.498 6.546 373 8.06 5.20 0.463 0.024 0.7% 1.0% 0.0% 3.6% 3.8%
7 6.985 7.071 376 8.99 4.88 0.489 0.025 1.2% 0.9% 0.0% 3.6% 3.9%
7.5 7.491 7.604 280 9.11 6.49 0.471 0.028 1.5% 0.9% 0.0% 3.6% 4.0%
8 7.995 8.108 251 10.01 6.59 0.537 0.034 1.4% 0.9% 0.0% 3.6% 4.0%
8.5 8.500 8.657 223 10.61 6.60 0.621 0.042 1.9% 0.8% 0.0% 3.6% 4.1%
9 8.991 9.145 185 10.95 7.54 0.595 0.044 1.7% 0.8% 0.0% 3.6% 4.1%
9.5 9.518 9.689 176 11.60 8.08 0.650 0.049 1.8% 0.8% 0.0% 3.6% 4.1%
10 9.972 10.113 143 12.19 8.69 0.596 0.050 1.4% 0.8% 0.0% 3.6% 3.9%
10.5 10.483 10.721 117 13.00 8.97 0.759 0.070 2.3% 0.8% 0.0% 3.6% 4.3%
11 10.989 11.100 64 14.13 9.35 0.785 0.098 1.0% 0.8% 0.0% 3.6% 3.9%
11.5 11.504 11.736 69 13.60 9.61 0.829 0.100 2.0% 0.8% 0.0% 3.6% 4.3%
12 12.012 12.272 72 14.34 10.74 0.824 0.097 2.2% 0.7% 0.0% 3.6% 4.3%
12.5 12.484 12.678 55 14.69 10.73 0.781 0.105 1.6% 0.7% 0.0% 3.6% 4.0%
13 12.997 12.835 31 14.84 11.76 0.611 0.110 -1.2% 0.7% 0.0% 3.6% 3.9%
13.5 13.541 13.408 23 14.26 11.91 0.541 0.113 -1.0% 0.7% 0.0% 3.6% 3.9%
14 13.992 13.743 25 14.79 12.86 0.557 0.111 -1.8% 0.7% 0.0% 3.6% 4.1%
14.5 14.510 14.432 16 16.29 13.61 0.605 0.151 -0.5% 0.7% 0.0% 3.6% 3.8%
15 14.913 14.870 11 15.83 14.03 0.587 0.177 -0.3% 0.7% 0.0% 3.6% 3.8%
15.5 15.500 15.041 13 15.99 14.25 0.620 0.172 -3.0% 0.7% 0.0% 3.6% 4.8%
16 16.052 15.339 12 17.46 14.23 0.957 0.276 -4.4% 0.7% 0.0% 3.6% 6.0%
ECOFYS WTTS B.V. | Kanaalweg 15G | 3526 KL Utrecht| T +31 (0)30 662-3827 | E info@ecofyswtts.com | I www.ecofyswtts.com
ECOFYS WTTS B.V.
Kanaalweg 15G
3526 KL Utrecht
T: +31 (0) 30 662-3827
E: info@ecofyswtts.com
I: www.ecofyswtts.com
top related