PRELIMINARY RESULTS Reference OF THE QA/QC Laboratory for - - PowerPoint PPT Presentation

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 1

• C. A. Belis, F. Lagler, L. Emblico and A. Borowiak

European Commission, Joint Research Centre IES, ERLAP

PRELIMINARY RESULTS OF THE QA/QC PROGRAMME FOR PM10/PM2.5 IN EUROPE

European Reference Laboratory for Air Pollution

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 2

CEN European PM standard: Why?

Standard Method to serve for purpose of EU - AQD => Harmonisation PM measurements in EU => Improve Data Quality PM measurements

Different implementations of the standard method

• r

different users of the standard method same/ comparable/ equivalent results!

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 3

• No primary standard feasible
• By convention :

Prescribed Standard Instrument, to serve as => “highest standards”

• not necessarily best metrological one –

– PM Standard Sampler dimensional properties – sampler materials – filter type(s) – sample flow rates – procedures for filter conditioning – weighing procedures CEN: European PM standard

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 4

CEN: the standards The standards: EN 12341 (1999): 3 reference methods (LVS, HVS, WRAC) + equivalence test method EN 14907 (2005): High Vol method (30 m3/h) and Low Vol method (2.3 m3/h) Equivalence guideline: http://ec.europa.eu/environment/air/pdf/test_equival encev31004.xls “Demonstration of Equivalence of ambient air monitoring methods”

• under revision-

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 5

Objectives:

• To provide information on the comparability of PM10 measurements

as implemented by the NRLs with those of JRC (ERLAP)

• To investigate the comparability of routine PM10 measurements at

network monitoring stations with those of JRC

• To assess, in the field, the comparability of reference and equivalent

methods and the achievement of the data quality objectives (DQO)

• To assess the state of implementation and use of correction factors

for automatic monitors in the monitoring networks that are used in reporting under Directive 2008/50/EC JRC – AQUILA PM QA/QC programme

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 6

• ne urban background site in each visited country
• parallel measurements between

JRC, National Reference Laboratory and Network monitoring stations/Institution performing routine assessment

• two weeks of parallel measurements (14 samples)

JRC – AQUILA PM QA/QC programme

Methodology:

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 7

Network monitoring station / NRL / institute responsible for PM correction factors in Member States: Minimum requirement: PM10 reference

parallel measurements

• Two reference samplers PM10
• Reference sampler PM2.5
• Sampler PM1
• OPC (size range 0.3 – 20 µm)
• TEOM FDMS (PM10)
• EC/OC semi-continuous
• Meteo (T, P, wind)

JRC – AQUILA PM QA/QC programme

Methodology:

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 8

2006 1. Spain - Madrid 2. Portugal - Lisbon 3. Slovenia - Ljubljana 4. Austria - Graz 5. Czech Republic - Prague 6. Germany - Wiesbaden 2009

• 17. Hungary - Budapest
• 18. Slovakia - Bratislava

2008

• 11. France - Paris
• 12. The Netherlands – Biest
• 13. United Kingdom – Port Talbot
• 14. Ireland - Dublin
• 15. Belgium - Antwerp
• 16. Italy - Milan

2007 7. Finland - Helsinki 8. Denmark - Copenhagen 9. Sweden - Stockholm

• 10. Estonia - Tallinn

JRC – AQUILA PM QA/QC programme

18 MS/ 27 MS

Sites:

not included in this evaluation

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 9

Sites visited during the PM QA/QC 2006 - 2009 Urban background Traffic Rural

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 10

• 2. Continuous PM10 monitor maintenance
• Cleaning of sampling head
• Cleaning of sampling tubes and switching valve
• Cross-check of built-in temperature and pressure sensors to certified

sensors

• Pump verification
• Verification / exchange of inline filters
• Verification of the calibration constant (K0)
• 1. Flow check gravimetric samplers

Requirements of EN14907 for PM2.5 applied to all samplers Tolerance: ± 2% of nominal flow QA/QC programme for PM: procedures followed

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 11

• 3. Weighing procedures and filter conditioning
• 1 µg balance resolution annually certified
• Climate-controlled weighing room with temperature control at 20

± 1 °C and RH 50 ± 5 %

• Quartz fiber filters, visually checked for defects prior to use
• Check of laboratory balances with standard (± 10 µg) weights

before any weighing session

• Filter conditioning parameters properly observed:
• conditioned for 48 hours
• filters weighed twice within 24 h interval

QA/QC programme for PM: procedures followed

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 12

PM10 Basic Statistics Me dia n; Box: 25% -75 %; Whisker: m in-ma x Spain Portugal Slovenia Austria Czech Rep. Germany Denmark Sweden Finland Estonia Netherlands Belgium France Ireland United Kingdom Italy 20 40 60 80 100 120 140 160 180 µg m-3 M e d i a n 2 5 %

• 7

5 % R a n g e

Concentration range for all campaigns

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 13

Robustness of the reference value Uncertainties for PM10 –JRC samplers A and B: (Calculation in accordance with EN14907, 9.3, eq.7)

) 100 / * / (

2 2 2 2 2 2 f m field c

u C V u u u + + =

μg/m3

0.1 Milan 0.33 Port Talbot 0.15 Dublin 0.14 Paris 0.72 Borgerhout 0.36 Biest - Houtakker 0.30 Tallinn 1.73 Helsinki 7.44 Stockholm 4.08 Copenhagen 0.54 Wiesbaden 0.09 Prague 0.99 Graz 0.06 Ljubljana 0.64 Lisbon 0.62 Madrid Between sampler uncertainty "ufield

2" in

μg/m3 Measurement site

ufield

2

Median: 0.45 μg/m3

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 14

Robustness of the reference value Entire JRC data set En – numbers: Data pairs: 230 Only 2 En-numbers >1 (1.06, 1.21) Median: 0.21

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 time µg/m3

Example: NL, B, F, IRE, UK

) 100 / * / (

2 2 2 2 2 2 f m field c

u C V u u u + + =

μg/m3

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 15

PM 2.5 and PM1 to PM10 ratio(%)

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0

M a d r i d L i s b

• n

L j u b l j a n a G r a z P r a g u e W i e s b a d e n C

• p

e n h a g e n S t

• c

k h

• l

m H e l s i n k i T a l l i n n B i e s t

• H
• u

t a k k e r B

• r

g e r h

• u

t P a r i s D u b l i n P

• r

t T a l b

• t

M i l a n

PM

2 . 5

/ PM

1

PM

1

/ PM

1

Avg., min, max (JRC)

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 16

Scatter plot ratios PM2.5 / PM1

y = 0.8238x - 0.0411 R2 = 0.9179 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.2 0.4 0.6 0.8 1 PM 2.5 PM 1

PM 2.5 and PM1 to PM10 ratio(%)

JRC

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 17

PM 10 all labs (y: LV / p: HV / b: online)

• 100
• 50

50 100 150 24 h average over time % deviation

All participants: PM10 compliance with DQO

PM10 all data 932 daily avg. n > DQO = 98 (10.5%) PM10 > “LV” (37.5 µg/m3) 270 daily avg. n > DQO = 22 (8.2%) PM2.5 all data 334 daily avg. n > DQO = 83 (24.9%) PM2.5 > ”LV” (18.75 µg/m3) 146 daily avg. n > DQO = 31 (21.2%)

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 18

PM10 grouped by sampler & filter type

PM10 - Low Vol Samplers

• 100
• 50

50 100 150

all 24 h averages over time % deviation from target

upper limit low er limit Zambelli PTFE Sw am Quartz Sw am Quartz Seq Leckel Quartz Seq Leckel Quartz Seq Leckel Glass Seq Leckel PTFE IVL unknow n IVL unknow n FH95KF nitrocell FH95KF glass Seq Derenda LVS glass Seq Derenda LVS quartz Derenda MVS 6.1 quartz Derenda MVS 6.1 quartz Tecora quartz Tecora unknow n Tecora PTFE Derenda 3.1-15(LVS) quartz Partisol PTFE Partisol PTFE Partisol glass Partisol quartz

filter type: quartz , glass fiber, − PTFE, cellulose nitrate

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 19

PM2.5 - Automatic analyzers

• 80
• 60
• 40
• 20

20 40 60 24 h averages over time % d e v ia tio n fro m ta rg e t upper limit lower limit FH62IR 30 deg Teom 50 deg Teom FDMS 30 deg SM200 40 deg Swam no heating Swam no heating Grimm 180 no heating FH62IR 40 deg FH62IR dynamic heating

PM2.5 automatic analyzers / sampling line temperature

30° C 40° C dyn heat

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 20

Application of PM10 Correction factors Applied factors:

• 13 out of 34 online datasets

applied a correction factor

• 10 FDMS, 2 Teom, 2 SM200,

2 FAI, 2 FH62IR, 1 Sharp, 1 Grimm, 1 MP 101 without a factor

• Either only factor or factor +
• ffset
• Range from 1.1 to 1.47

Consequences of applied factors with respect to DQO: Out of 13 datasets (each 12 – 14 daily averages):

• 9 improved
• 3 worsened
• 1 unchanged

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 21

PM10 / PM2.5 data expressed at standard conditions The volume flow rate should be expressed at ambient conditions (EN 14907)

• 6 laboratories (PM10) and 2 laboratories (PM2.5) expressed

data at conditions different from ambient

• The applied standard conditions were:

1013 hPa / 0 °C 1013 hPa / 20 °C 1013 hPa / 25 °C

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 22

PM10 uncertainty

• 12 laboratories (number might still increase) out of 31 laboratories

expressed uncertainties

• Calculation of En-number (ISO/IEC guide 43-1):

Example for LANUV (D) at the annual limit value u (LANUV) = 2.25 µg/m3 u (JRC) ≈ 1.3 µg/m3 In this comparison were included only values >30 µg/m3 (= annual limit value of 40 µg/m3 - 25%)

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 23

PM10 uncertainty JRC samplers A and B vs. LANUV samplers 4 and 37 6 daily averages >30 µg/m3 0.07 0.07 0.14 0.14 0.01 0.21 0.21 0.01 0.62 0.69 0.28 0.34 0.25 0.09 0.34 0.19 0.08 0.11 0.01 0.02 0.01 0.06 0.01 0.05 En (B/37) En (A/37) En (B/4) En (A/4)

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 24

PM10 uncertainty

• Calculation of uncertainty at the annual limit value (example LANUV):

Raw Data 5 10 15 20 25 30 35 40 45 50 10 20 30 40 50

R eference M etho d

ug/m3 0.74 RM between-sampler uncertainty pass 2.27 relative uncertainty at the LV ug/m3 0.91 combined uncertainty ug/m3

• 0.26

bias at LV ug/m3 0.87 random term EQUIVALENCE TEST RESULTS 0.72 uncertainty of a not significant

• 1.35

intercept a 0.02 uncertainty of b not significant 1.03 slope b REGRESSION OUTPUT UNCORRECTED DATA

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 25

PM2.5 uncertainty and “Average Exposure Index” Possible influence of PM2.5 uncertainty on “Reduction target”

JRC - PM2.5 concentrations 8.5 < PM < 13 µg/m3

6 7 8 9 10 11 12 13 14 15 16

time µg/m3

reduction target + reduction target -

• Concentr. + U in μg/m3

But: these are daily averages while AEI are 3 year averages!

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 26

Average PM10 compositions: contribution of natural sources can account for up to 50% of PM mass

PM10 Lisbon

nssSO4 6% NH4 1% nssCa 2% Other ions 0% EC 5% NO3 7% Sea salt 43% OM 7% Unacc. 29%

PM10 Madrid

nssSO4 5% NH4 1% EC 11% Other ions 0% Unacc. 56% OM 18% Sea salt 0% NO3 4% nssCa 5% y = 9.3x R2 = 1.0 10 20 30 40 50 2 4 6 nssCa (µgm-³) Unaccounted (µgm-³)

In Madrid the water insoluble inorganic fraction (Unacc.) is in very good correlation with nssCa, a tracer for crustal matter Chemical composition of PM10: Lisbon, Madrid

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 27

Field blanks

• EN14907: “…if the blank (field) masses differ by more than 40 µg for LVS

…… the reason shall be investigated.”

• 62 filed blanks out of 110 (56%) differed more than 40 µg
• Worst case Budapest and Bratislava with ca. 300 µg !!!
• Max. influence on mass concentration:

PM10: 49.6 % (Bratislava) PM2.5: 62.1 % (Bratislava)

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 28

Field blanks HU and SK

• Investigations (HU and SK):
• Heating a “loaded” field blank to 30 °C for 24 h removed most of the

gained mass semivolatile compounds

• Confirmed by

comparison to FDMS e.g. HU: Field blank accounts for 5.8 µg/m3):

Budapest FDMS vs. SEQ y = 0.9814x + 5.9302 R2 = 0.9938 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 0.00 20.00 40.00 60.00 80.00 100.00 FDMS Gravimetry

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 29

Concluding remarks

• The uncertainty between two identical parallel PM10 sampling systems

used as reference was negligible.

• There is a good agreement between reference and NRLs for gravimetric
• methods. The highest discrepancies are observed in the extremes of the

mass range (very high or very low concentrations).

• PM10 measurements comply with DQO in most cases while uncertainty for

PM2.5 measurements is higher than DQO in >20 % of the reported data.

• At 50% of the L.V., the uncertainty of PM2.5 is likely to be comparable to the

reduction targets set by EU AQD for that level (needs further research).

• Differences are more evident in the comparison of reference methods with

continuous monitors. This is partly due to the variety of operation principles and sample treatment (e.g. heating).

• For this type of measurements the application of correction factors

contributes, in the majority of cases, to achieve DQO.

• The loads in travelling blanks are not negligible. Work is been carried out to

identify the causes and to find out technical solutions.

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 30

Thanks to the ERLAP team

Annette Borowiak, Lorenza Emblico, Michel Gerboles, Agnieszka Kupc, Fritz Lagler, Luisa Marelli, Julien Theunis

and the NRLs & local Networks!

AEA Technology Hessisches Landesamt fuer Umwelt und Geologie Agenzia Regionale per la Protezione dell'Ambiente della Lombardia Ineris AIRPARIF Instituto Carlos III Amt der Oberoesterreichischen Landesregierung Instituto do Ambiente Amt der Steiermaerkischen Landesregierung ISSeP, Veille Technologique, Cellule qualitè de l'Air ANHOVO Istituto Superiore per la Protezione e Ricerca Ambientale Ayuntamiento de Madrid IVL Swedish Environmental Research Institute Bureau Veritas Landesamt fuer Natur, Umwelt und Verbraucherschutz City of Stockholm Environment and Health Administration LCSQA - Mines de Douai Département Chimie & Environnement Czech Hydrometrological Institute National Environmental Research Institute Dublin City Council Neath Port Talbot Borough Council and Air Monitors Elektroinstitut Milan Vidmar EIMV - department ENV RIVM - National Institute for Public Health and the Environment Environmental Agency of the Republic of Slovenia Umweltbundesamt Ges.m.b.H Environmental Protection Agency University of Helsinki Estonian Envirionmental Research Centre University of Stockholm ITM Department of Applied Environmental Science Finnish Metrological Institute VMM - Vlaamse Milieumaatschappij Afdeling Lucht, Milieu en Communicatie GGD - Municipal Health Service Amsterdam

14th EIONET WORKSHOP on AIR QUALITY ASSESMENT AND MANAGEMENT, 5-6th October 2009 31

Network Members: 37 National Reference Laboratories from 27 EU Member States + Switzerland, Norway. Co-ordinated by JRC (ERLAP). Chaired by JRC, DG ENV and elected NRLs.

http://ies.jrc.ec.europa.eu/aquila-homepage.html

→ Promote the harmonisation of AQ measurements in EU & CC → Co-ordinate and participate to QA/QC activities, method development and validation

AQUILA Network