assessment of low-cost sensors for air quality in real-world ......pod (aqmesh): pm 1, pm 2.5 and pm...
TRANSCRIPT
Assessment of low-cost sensors for
air quality in real-world conditions
María Cruz Minguillón, Mar Viana, Cristina Reche, Fulvio Amato, Xavier Querol
Meeting on sensor technology for air quality Bilthoven, 13 February 2017
Institute of Environmental Assessment and Water Research (IDAEA), CSIC
What is a low-cost air quality sensor?
An instrument to measure air quality that…
- is small
- is low-cost
- can be deployed outdoors or indoors
- can be deployed for a reasonable amount of time
• They are complementary to air quality networks
• Laboratory performance is satisfactory, although their validation in real-world
conditions still has limitations
cheap enough to be replaced rather than fixed if it fails (Morawska, 2016)
<1 kg?
1 month?
• OBJETIVE: assess the performance of air quality sensors in real-world
conditions
Bilthoven, 13 February 2017
Methodology
Urban background site in Barcelona, with pollutant concentrations
typical of the urban background in the Mediterranean region
Bilthoven, 13 February 2017
Palau Reial, IDAEA-CSIC
PM: Optical particle counter (GRIMM 180)
O3: Ultraviolet absorption SIR S-5014
NO and NO2: Chemiluminescence SIR S-5012
PM: High volume captor and gravimetric determination
Bilthoven, 13 February 2017
Methodology
ELM / CANARIT O3 Metal oxide
POD (AQMesh) NO, NO2, O3 Electrochemical
PerkinElmer (before AirBase)
CAPTOR O3 Metal oxide
Particulate matter
DYLOS
MICROPEM AIRBEAM
Tested sensors
POD (AQMesh)
Bilthoven, 13 February 2017
Gaseous pollutants
X stopped
Pod (AQMesh): NO, NO2, O3
y = 0.69x - 6.9R² = 0.90
0
50
100
150
200
250
300
0 200 400
NO
(µ
g/m
3)
AQ
Me
sh
NO (µg/m3) Reference
y = 0.82x + 1.1R² = 0.99
0
100
200
300
400
0 200 400
NO
(µ
g/m
3)
AQ
Me
sh
NO (µg/m3) Reference
Modification
Equipped with a solar roof to reduce temperature effect
Bilthoven, 13 February 2017
Pod (AQMesh): NO, NO2, O3
Modification
Improve in technology
and data treatment
results in an increased R2
Bilthoven, 13 February 2017
Pod (AQMesh): NO, NO2, O3
Modification
ATTENTION!
They can vary over time.
Worse correlations in
summer than winter
R2 = 0.72
Improve in technology
and data treatment
results in an increased R2
Bilthoven, 13 February 2017
Captor: O3
H2020 project
Bilthoven, 13 February 2017
Inter instrument
comparison
Captor: O3
H2020 project
Bilthoven, 13 February 2017
Field deployment
Example 1
Captor: O3
H2020 project
Bilthoven, 13 February 2017
y = 0.588x + 10.881R² = 0.4715
0
20
40
60
80
100
120
140
160
180
200
0 50 100 150 200
Cap
tor
O3
(µg/
m3)
Reference O3 (µg/m3)
Field deployment
Example 2
Dylos: N>0.5 and N>2.5
• 3 units
• >1 year
Bilthoven, 13 February 2017
• N0.5-2.5 calculated
• Proxy for PM2.5
• Time res: 1min
• Averaged to
5min or
30min
Inter instrument comparison
2 months
Dylos: N>0.5 and N>2.5
Bilthoven, 13 February 2017
y = 16859x - 144570R² = 0.7047
0
100000
200000
300000
400000
500000
600000
0.0 10.0 20.0 30.0 40.0D
ylo
s_1
(#/
ft3)
Ref PM (µg/m3)
d1_aprmay
RH= 67,74%T=17,18°C
y = 29573x - 305052R² = 0.7052
0
200000
400000
600000
800000
1000000
1200000
1400000
1600000
0.0 10.0 20.0 30.0 40.0 50.0
Dyl
os
(#/f
t3)
Ref PM (µg/m3)
d1_mar
RH= 64,13%T=12,69°C
Comparison with reference
y = 38856x - 373608R² = 0.8546
0
200000
400000
600000
800000
1000000
1200000
1400000
1600000
0.0 10.0 20.0 30.0 40.0 50.0
Dyl
os_
1 (
#/ft
3)
Ref PM (µg/m3)
d1_janfeb
RH= 65,21%T=13,64°C
Pod (AQMesh): PM1, PM2.5 and PM10
Bilthoven, 13 February 2017
• Time res:
1 min every 15 min
• Averaged to 1h
• 1 unit
• 5.5 months
0
20
40
60
80
100
17/01/2016 06/02/2016 26/02/2016 17/03/2016 06/04/2016 26/04/2016 16/05/2016 05/06/2016 25/06/2016
POD_PM2.5
REF_PM2.5
0
25
50
75
100
125
150
175
17/01/2016 06/02/2016 26/02/2016 17/03/2016 06/04/2016 26/04/2016 16/05/2016 05/06/2016 25/06/2016
Pod_PM10
REF_PM10Comparison with
reference
Pod (AQMesh): PM1, PM2.5 and PM10
Bilthoven, 13 February 2017
• Time res:
1min every 15 min
• 3 units
• 1 month
PM10
PM2.5
PM1
Inter instrument
comparison
Pod (AQMesh): PM1, PM2.5 and PM10
Bilthoven, 13 February 2017
• Time res:
1min every 15 min
Matched to 5min reference average PM10
PM2.5
PM1
Comparison with reference
• 3 units
• 1 month
Airbeam: PM2.5
Bilthoven, 13 February 2017
0
10
20
30
40
08/08/2016 0:00 13/08/2016 0:00 18/08/2016 0:00 23/08/2016 0:00 28/08/2016 0:00 02/09/2016 0:00
Airbeam_1 Airbeam_2
• PM2.5
• Time res: 5min
• Averaged to 30min
• 1 month
Inter instrument comparison
Airbeam: PM2.5
Bilthoven, 13 February 2017
• PM2.5
• Time res: 5min
• Averaged to 30min
• 3 months
Comparison with reference
0
5
10
15
20
25
30
35
40
45
50
09
/08
/16
14
/08
/16
19
/08
/16
24
/08
/16
29
/08
/16
03
/09
/16
08
/09
/16
13
/09
/16
18
/09
/16
23
/09
/16
28
/09
/16
03
/10
/16
08
/10
/16
13
/10
/16
18
/10
/16
23
/10
/16
PM
2.5
(μg
m-3
)
Airbeam IDAEA
Reference
Airbeam: PM2.5
Bilthoven, 13 February 2017
• PM2.5
• Time res: 5min
• Averaged to 30min
• 3 months
Comparison with reference
Airbeam: PM2.5
Bilthoven, 13 February 2017
Application:
Awareness raising,
teaching in schools
MicroPEM: N<2.5 or N<10
Bilthoven, 13 February 2017
Inter instrument comparison
MicroPEM: N<2.5 or N<10
Bilthoven, 13 February 2017
0
20
40
60
80
10/mar 15/mar 20/mar 25/mar 30/mar
MicroPEM PM10
Reference PM10
Comparison
with reference
Intercomparison exercise 1st EuNetAir Air Quality Joint Intercomparison Exercise Aveiro (Portugal) 13-27 October 2014
Aveiro
Measured parameters: • CO, NOx, O3, SO2, • PM10, PM2.5, • temperature, relative
humidity
IDAD-Institute of Environment and Development Air Quality Mobile Laboratory
Bilthoven, 13 February 2017
- 15 participating teams - 130 microsensors - 27 working sensors
Conclusions
Some nodes provide satisfactory or acceptable results
Usually good agreement between different nodes of the same type
Changes over time: required frequent validation with reference
instrumentation
Influence of temperature for gaseous pollutants measurements
Optical particle counters valid as proxy for fine particles (<2.5 µm)
Most particle nodes not adapted for outdoor deployment
More tests and improvement of technology and data processing required
The possibility to use sensors for air quality assessment can be a fact provided
some improvements and validations are applied
Bilthoven, 13 February 2017
Thank you for
your attention
Acknowledgements
Anna Ripoll, Marc Padrosa, Francesco Miacola, Giorgos Spais