ICE-FREE SENSORS - THE EUMETNET SWS II PROJECT
Bengt Tammelin*, Alain Heimo** and Michel Leroy***
*Finnish Meteorological Institute, ** Meteo Swiss, *** Meteo France
bengt.tammelin@fmi.fi
ABSTRACT
Wind, temperature, humidity and solar sensors were tested during the winter 2001/02 under heavy atmospheric
icing conditions at three test sites located in southern France, Switzerland and northern Finland. There are some
significant differences between the sensors in respect to operation during icing periods. Some of the sensors operate
very successfully under harsh conditions, as some can be improved by improved heating or protection against ice
accretion. In this paper the EUMETNET SWS II project is briefly described. Also some examples of analyses of wind
sensor data are shown.
1
BACKGROUND
Ice and rime accretion upon meteorological sensors due to atmospheric icing is a major problem for accuracy of
measurements done in cold climate conditions at high latitudes and high altitudes. Most of present sensors operated e.g.
by the national weather services are not ice free sensors, even if they may have some partial heating. The need for icefree sensors is increasing not only due to meteorological or synoptic purposes, but also due to requirements on accurate
measurements at hostile environments concerning applications linked e.g. to structural design, exploitation of
renewable energy sources, aviation, power lines, skiing centres etc.
Concerning accuracy and reliability of measurements the requirements introduced by the WMO [ 3] are followed
widely. Unfortunately the effect of atmospheric icing is not discussed in this guide. However, traditionally e.g. axial
heating or black painted cups for wind sensors [ 1] and special precipitation gauges have been used at sites located in
cold climate regions.
Finnish Meteorological Institute has studied icing effect and ice-free sensors especially concerning wind measurements
(e.g. [ 1],[ 2], [ 4]). Also other reported experiments/observations on icing effects on sensors are available. The
EUMETNET1 project "Specification of Severe Weather Sensors 1997-1998" summarised the icing effects on different
types of meteorological sensors [ 5]. Following the first SWS project the SWS II project was started by the
EUMETNET in July 2000 in order to test a number of ice-free sensors, and other measurement arrangements designed
for cold climate conditions, at three different types of sites in Finland, France and Switzerland. The final report of the
project is expected to be ready by the end of the year 2002.
The objective of the project is to systematically test and analyse ice-free instruments (humidity, temperature, wind
speed, wind direction, solar radiation) under harsh icing conditions to produce data for representative scientific studies:
- for intercomparison of ice-free sensors used or preferred by EUMETNET members
- to produce information on errors and the amount of heat needed for sensors to be ice-free
- to start a network providing certification for sensors that are free of icing problems.
This paper describes the SWS II project and some preliminary results concerning wind sensors. Temperature, humidity
and solar radiation measurements at these sites will be discussed in three other papers presented at the TECO -2002
(see: [ 8], [ 9], [ 10]).
2
TEST SITES
The three test are located in different parts of Europe under various types of cold climate and icing conditions. All these
sites are within areas of heavy or strong atmospheric icing (with number of icing days more than 15 d/y [ 6]) and high
wind speeds. However, in the far north the winter days are short with very little solar radiation to melt the ice, while
close to the Mediterranean daily solar energy significantly reduces the amount of ice upon sensors (depending e.g. on
the size, shape and colour of the sensors) but also affects the structure of ice. One reason to have the test sites at such
places is to have enough events with atmospheric icing for representative analyses of icing effects upon sensors and
measurements
The Mont Aigoual station, which is part of the basic network of Météo-France, is located on the top of Mont Aigoual
(altitude 1567 m a.s.l., 44º07´N, 03º35´E). The MeteoSwiss test station at Säntis mountain is close to the MeteoSwiss
1
EUMETNET is a network of 18 National Meteorological Services: those of the EU plus Iceland, Norway and
Switzerland; www.eumetnet.eu.or
weather station LPNN 2220. The test platform is at altitude of 2490 m a.s.l. (47º15´N, 9º20´E). The FMI test station is
located in northern Finland north of the Polar circle on the top of Luosto fell at 515 m a.s.l. (68º08´N, 26º54´E;
LPNN7509/WMO 05841). Luosto is at the northern end of a chain of arctic fells. At each site a platform for sensors
was built (Figure 1). Each test site has an automatic data acquisition system providing with 10 minute averages and 3 s
min an max values and standard deviations. Mont Aigoual and Säntis are manned stations where the observers also
frequently take photos of the status of the sensors. Luosto is a fully automatic station where the status of sensors is
monitored by three properly heated (ice-free) video cameras (for more detailed description of the sites see [ 7]).
Figure 1. Test platforms with different sensors at Mont Aigoual (left) and Säntis (right)
Figure 2. Examples of video monitoring at Luosto: on the left the rotating Cam2 watching wind sensors, and on the
right the fixed Cam1 watching the ice detectors. Wind sensors in the Cam2 image from the left:Thies, Lamrecht, Metek
55W, Vaisala sonic, Rosemount, Vaisala cups and Irdam.
3
SENSORS TESTED
The sensors to be tested during the winter 2001/02 at various sites were chosen according to interest of the
EUMETNET members, and the discussions had at the SWS II workshop held in Paris in June 2001. Thus e.g. some
wind sensors known to be ice free but not able to meet the WMO requirements concerning accuracy of measurements
especially at mountainous sites where e.g. vertical wind component affects strongly the measurements [ 4] were
excluded.
Measurements of icing is not actually included in the SWS II project, but as information of rime/ice accretion is
required for other analyses also some ice-detectors and observation methods were tested.
According to results from the test period 2000/2001 it was decided that ice/rime upon sensors was not to be removed by
observers during the test period. Thus all sensors were performing like at an automatic weather station. The sensors
tested at Luosto (L), Säntis (S) and Mont Aigoual (A) were:
A) wind sensors: R.M. Young Wind Monitor-Model 05103-45 (A), Irdam SA WST 7000HS (A,L), Metek USA-1 55
W (L), METEK USA-1 125 W (A, S), Degreane Deolia 300 (A), Gill Windobserver II (S), Firma Kroneis Zamg
263PRH (A), Theodor Friedrichs Type 4035.0000 and Type 4123.0000 (L), A. Thies Ultrasonic 2D (L), Lambrecht
Static Wind Sensor M16420 (A, L), BF Goodrich Rosemount M1774W (A, L, S), Meteolabor NOWA (S), FT
Technology FT702 (S), Vaisala WAA252 (L, S), Vaisala WAS425 (A, L, S),
B) temperature and humidity: Rotronic ventilated radiation shield (A, L, S), Meteolabor Thygan (A, L, S), Eigenbrodt
ventilated sheild with HMP45 (L), and additionally Irdam WST7000HS (A, L), Metek USA-1(A, L, S), GILL
Windobserver (S), NOWA (S),
C) solar radiation: Kipp&Zonen CM21 with MeteoSwiss ventilation, cleaned and uncleaned (S), CM11 with FMI
ventilation system (S, L), Hänni solarimeters (S),
D) icing: Labko LID-3503 (L, S), BF Goodrich Rosemount ice detector (A, L, S), Vaisala FD12P (L).
The reference sensors at all three sites were chosen to be Rosemount Model 1774W for wind speed and wind direction,
and THYGAN for temperature and humidity measurements, as these were seen to operate most correctly at all three
sites also during heavy icing. All other sensors were verified to these sensors also using data from non-icing periods.
4
PRELIMINARY RESULTS - WIND MEASUREMENTS
Number of icing days, intensity of ice accretion and type of icing varies between the three sites. In South solar radiation
helps to remove the ice, but also turns rime to clear ice due to melting and icing, while in North solar radiation has no
effect in November-February.
Figure 3. An example of definition of
icing classes for some wind sensors
(WAS425 above and Young below). Class
1 on the left where there is some ice upon
the body and sensor, but the sensor head
is free of ice, Class 2 in the middle where
ice is detected also upon the sensor head,
and on the right Class 3 where the
sensor is totally blocked. For WAS425
one of the sensor heads at each picture is
shown by a circle.
According to the amount of ice upon sensors (body, shaft, cups etc.) the severity of icing upon each sensor is divided
into four classes: 0 = totally clear, 3 = totally blocked. At Luosto these analyses were based on continuous video
monitoring. At Säntis the classification was done by the observer, and at Mont Aigoual based on photos taken by the
observer
To verify icing effect on wind sensors only open (no obstacles or shade of neighbour instrument) wind sectors were
used. To study duration of different ice amounts on sensors all data was used.
In general availability of representative data measured with the "3 rd generation2 " heated ice-free sensors is much higher
than with ordinary shaft heated sensors. Thus heated wind sensors available offer improved opportunities for proper
wind measurements even at harsh climates. However, there are significant differences between the wind sensors in
respect to icing effects on measurements. According to the results from Luosto in Figure 5 Thies, Lambrecht, Irdam and
Rosemount are practically ice free all the time, while WAA252 and Metek 55 W require some additional heating. For
Thies and Rosemount class 1 (ice upon the body) does not really affect the wind measurements. On the other hand,
Irdam and Lambrecht sensors seem to be very sensitive to water droplets carried by hard wind into the sensor, which
causes strong overestimation of wind speed. WAA252 is partially affected by strong vertical wind component, which is
typical for these sites, and which leads to overestimation of wind speed (see Figure 4 on the right).
2
1st generation: external heating elements in early 1980s, 2 nd generation: heavy internal heating since middle of 1980s,
3rd generation: direct low power heating of sensing parts of the sensor since middle of 1990s.
SWS Luosto, February 2002
THE_ff vs ROS_ff ( sector 160 - 240 deg., 10 min avg)
SWS Luosto, February 2002
VAI_ff vs ROS_ff ( sector 160 - 240 deg., 10 min avg)
30
30
N =1581
N0 = 0
N1 = 43
N2 = 197
N3 = 301
25
Class 0
25
Class 1
20
y = 1,0585x - 0,159
Class 0
Class 1
20
Class 2
Class 2
Class 3
15
Ref
m/s
m/s
N =1581
N0 = 0
N1 = 560
N2 = 0
N3 = 0
y = 1,0949x - 0,4551
Class 3
15
Ref
10
10
5
5
0
0
0
5
10
15
20
25
0
5
m/s
10
15
20
25
m/s
Figure 4. An example of verification of wind sensors (Friedrichs on the left, WAA252 on the right) to the Rosemount at
Luosto in February 2002. The regression line describes the relation between the sensors during non icing period. On
the left all icing classes were detected, on the right only the class 1 is observed. Each dot represents 10 minutes average
wind speed.
Class 0
Class 1
Class 2
Class 3
SWS Luosto, Feb-Mar 2002
Duration of Icing (%)
100
80
%
60
40
20
0
THE
THI
LAM
MET
HAN
VAI
IRD
Figure 5. Duration (% of time) of
different types of ice classes
(class 0 on the left and class 3 on
the right) upon different wind
sensors at Luosto in JanuaryFebruary 2002. Sensors from
left:
Friedrichs,
Thies,
Lambrecht, Metek 50 W, Handar
or WAS425, WAA252, Irdam,
Rosemount.
As seen for
Rosemount the class 1 is
dominant (ice upon the body),
which, however, does not actually
affect the measurements.
ROS
Wind sensors
5
REFERENCES
[ 1] Tammelin, B., 1982. Frost formation on anemometers and frost prevention experiments. Technical report No 26.
Finnish Meteorological Institute, Helsinki. 34 p.
[ 2] Tammelin, B., 1992. Experiences of wind measurements on fell peaks. In the proceedings of BOREAS conference
1992 (Ed. B. Tammelin et al.). Finnish Meteorological Institute, Helsinki. p. 241-246.
[ 3] WMO, 1996. Guide to Meteorological Instruments and Methods of Observation. WMO-No. 8. Secretariat of the
World Meteorological Organization - Geneva - Switzerland.
[ 4] Tammelin, B., Hyvönen, R. and Peltomaa, A., 1996. The accuracy of wind measurements in hilly regions. In the
proceedings of BOREAS III 19-21 Saariselkä, Finland. (Ed. B. Tammelin et al.). Finnish Meteorological
Institute, Helsinki. p. 232-250.
[ 5] Tammelin, B., Joss, J. and Haapalainen, J., 1999. Specification of severe weather sensors - Final report of the
EUMETNET project. Finnish Meteorological Institute, Helsinki. 154 p. + Annexes.
[ 6] Tammelin, B., Cavaliere, M., Holttinen, H., Morgan, C., Seifert, H. and Säntti, K., 2000. Wind energy production
in cold climate. Meteorological Publications No 41. Finnish Meteorological Institute; Helsinki. 41 p.
[ 7] Tammelin, B., Heimo, A., Leroy, M., Rast, J. and Säntti, K., 2001. Meteorological measurements under icing
conditions - EUMETNET SWS II project. Reports 2001:6. Finnish Meteorological Institute. 52 p.
[ 8] Leroy, M. et al. Temperature/humidity measurements during icing conditions. To be presented at TECO 2002.
[ 9] Musa, M. et al. Measurements of temperature with wind-sensors during severe winter conditions. To be presented
at TECO 2002.
[ 10] Rast, J. et al. Solar radiation measurements under icing conditions. To be presented at TECO 2002.