Comparison between
temperature sensors
in plastic, aluminium shields
and “Stevenson” screen
Franco Stravisi, Stefano Cirilli
(University of Trieste, Dept. Mathematics and Geosciences)
TRIESTE: northern Adriatic Sea
The meteorological station is near the sea; University is on the hills.
The meteorological station is on the roof of the Nautical Institute, close to the sea.
The place is the same since 1819 (two centuries).
The meteorological instruments are on an airy wooden terrace over the roof of the
Nautical Institute. The large conventional screen and the test screens are showed.
LWS
TRIESTE station; large wood
shield (LWS) with double walls and
conventional instruments.
In 1992 a Pt100 was located inside
LWS for comparison with the small
PVC screen outside.
Micros datalogger (1992-2011)
and Pt100 (Class B) screen
NESA datalogger (since 2011)
Air temperature data at TRIESTE:
1. 1802 – 1883: thermometer readings, 3/day; MIN / MAX
2. 1884 – 2000: thermographs & MIN / MAX thermometers, conventional wood
shield
3. 1992 – 2011: Pt100 (*), small PVC screen, 10 min averages. MICROS
datalogger, modem link.
4. 2011 – present: Pt100 (**), small aluminum screen, 10 min averages. NESA
datalogger, web link.
(*) Pt100: Class B, 4-wire cable, calibrated in situ
(*) Pt100: 1/3 DIN, 4-wire cable, calibrated in situ
Necessary to:
A) check and ensure the continuity between conventional and “modern”
thermometers/screens;
B) test the performance of different current screens
(A) Conventional wood screen vs. small PVC screen
1996 – 2010
Trieste 1996-2010:
monthly mean temperature /°C
30
y = 1.000 x - 0.05
25
reference screen
LWS vs. SR
r2 = 1
20
(1)
15
Monthly average temperatures
measured by means of Pt100
sensors (in situ calibrated) inside
the large wood shield and in the
small PVC screen (SR) are the
same.
10
5
0
0
5
10
15
20
wood conventional screen
25
30
(10 min data from 1996 to 2010)
Conventional wood screen vs. small PVC screen
Trieste 1996-2010: monthly means of
daily minimum temperature /°C
LWS vs. SR
30
y = 0.997 x - 0.08
r 2 = 0.9999
reference screen
25
20
(2)
15
Monthly means of daily
minimum (*) temperatures
measured by means of Pt100
sensors (in situ calibrated)
inside the large wood shield
and in the small PVC screen
(SR) are the same.
10
5
0
0
5
10
15
20
wood conventional screen
25
30
(*) the minimum of 144 daily
10 min averages
Conventional wood screen vs. small PVC screen
Trieste 1996-2010: monthly means of
daily maximum temperature /°C
LWS vs. SR
35
y = 1.008 x - 0.03
r 2 = 0.9999
reference screen
30
25
(3)
20
Monthly means of daily
maximum (*) temperatures
15
measured by means of Pt100
sensors (in situ calibrated)
inside the large wood shield
and in the small PVC screen
(SR) are the same.
10
5
0
0
5
10
15
20
25
wood conventional screen
30
35
(*) the maximum of 144 daily
10 min averages
INITIAL CONSIDERATION:
After a long comparison between the conventional large wood screen LWS in use
at the meteorological station of Trieste and a small PVC screen (SR) during 15
years (1996 to 2010), with Pt100 probe on site calibrated and adapted ,we can
consider:
• these particular shelters equivalent;
• 10-min averaged daily extreme Pt100 temperatures are equivalent to
readings of standard MIN / MAX thermometers.
MAX
MIN
=
LWS = SR
LWS shows a time
delay ≤ 10 min with
respect to SR
(B) Comparison of different screens
SR reference screen,
PVC, 20 cm cones,
teflon stem
S1 all aluminium,
16 cm cones
S1’ one more open
bottom cone
S2 aluminium, PVC top
S3 same with 7 cones
S4 6 aluminium cones,
large PVC top,
teflon ring between
sensor and stem
REFERENCE SCREEN: our choice does
not mean that screen SR is the “best one”
in absolute, but just that it assures
continuity in our climatic temperature
series, performing like LWS.
Similar study on commercial screens of
different types have been made e.g. by
M. Lacombe, D. Bousri, M. Leroy, M. Mezred, “WMO
field intercomparison of thermometer screens/ shields
and humidity measuring instruments”, in: WMO
Instruments And Observing Methods Report No. 106.
Our purpose here, is to test some
different screens customized and made
available to us, by a company (Nesa
S.r.l.) who produces meteorological
equipment and datalogger..
In the following, attention is limited for
simplicity to screens S1 and S4; S1’, S2
and S3 behave in an intermediate way.
Test screens; the shadowed courtyard below prevents direct ascent of hot air.
From left: SR, S3, S4.
S1
0.39 ± 0.59 °C
Departure from SR
reference temperature
(overheating)
S4
0.15 ± 0.24 °C
4
4
S4
3
3
2
2
overheating /°C
overheating /°C
S1
1
0
-1
1
0
-1
-2
-2
0
2
4
6
8
10
12
14
16
10 min average wind velocity /(m/s)
18
20
0
4
2
4
6
8
10
10 min average wind velocity /(m/s)
14
4
S1: 0.4 ± 0.6 °C
S4: 0.2 ± 0.2 °C
3
overheating /°C
3
overheating /°C
12
2
1
0
2
1
0
-1
-1
-2
-2
0
200
400
600
irradiance /(W/m2)
800
1000
0
200
400
600
irradiance /(w/m2)
800
1000
Mean overheating (i.e. departure from the reference screen) diurnal cycle,
screens S1 and S4,
Screens S1’ to S3 present an intermediate behaviour.
1.0
S4 - SR
S1 - SR
overheating /°C
0.8
0.6
0.4
0.2
0.0
-0.2
0
6
12
time (CET) 18
24
S1
S3
S4
Examples of infrared pictures of
different screens
(see under the dish on the top of sensor)
RAIN PROTECTION
Screen S4 seems to be more sensitive than the larger SR to intense
rainfalls. Water could produce a temperature drop (psychrometric
effect).
34
14
32
12
30
10
28
8
26
6
SR
S4
precipitations
24
25.2 °C
4
22
2
21.5 °C
20
0
12
13
14
15
16
17
time (CET)
18
precipitations /mm
air temperature /°C
Example (16 July 2014):
Air temperature shelters/screens: final results
(A) Air temperature data measured in LWS with mechanical thermographs and
MIN/MAX reference thermometers are equivalent to 10 min averages data
logged by means of Pt100 in a small PVC screen calibrated and adapted (SR).
(B) Different sensor screens can show different degrees of overheating, increasing
with sun irradiance in still air.
In order to minimize that, the sensor should be white painted, the inner stem
thermally insulated from the supporting arm, the air should flow freely also
along the vertical; cones should cover a good part of the stem and a flat top
dish should shadow the whole screen.
Our results (comparing SR and S4, of similar shape) indicate that SR is
affected by just a little bit smaller overheating (0.2 °C in average) than S4,
substantially equivalents; further investigation could explain whether this
depends on the different material used or the size.
Larger screens give a better protection from intense rainfalls.
Acknowledgements: special thanks to NESA s.r.l. that patiently supplied us
many screens specially assembled according to our requests ...