Progress report from AQUILA Network of Air Quality Reference Laboratories and JRC - ERLAP

Transcription

1 EIONET Meeting Progress report from AQUILA Network of Air Quality Reference Laboratories and JRC - ERLAP Annette Borowiak Joint Research Centre (JRC) Institute for Environment and Sustainability (IES) Transport & Air Quality Unit -> Climate Change & Air Pollution Unit

2 EIONET Meeting Structure of presentation AQUILA and recent activities Geostatistics in uncertainty evaluation of air quality data Source apportionment intercomparison initiative Announcement of AirMonTech workshop

3 AQUILA EIONET Meeting Network Members: 37 National Reference Laboratories from the 27 Member States & EFTA Observers: Turkey, Croatia, Macedonia, Serbia annette.borowiak@jrc.ec.europa.eu

4 AQUILA: Members EIONET Meeting

5 AQUILA: associated members & observers EIONET Meeting World Health Organisation Collaborating Centre (Hans-Guido Muecke) European Environment Agency (Valentin Foltescu) European Topic Centre on Air and Climate Change (Frank De Leeuw) Institute of Public Health, Belgrade, Republic of Serbia Ministry of Environment and Physical Planning,Skopje, Republic of Macedonia Energy and Environmental Protection Institute (EKONERG), Zagreb, Croatia Ministry of Environment and Forestry, Ankara, Turkey

6 AQUILA: background EIONET Meeting Article 3 (2008/50/EC): Responsibilities For the implementation of this Directive, the Member States shall designate at the appropriate levels the competent authorities and bodies responsible for: Assessment of ambient air quality, Approval of measurement systems (methods, equipment, networks, laboratories), Ensuring accuracy of measurements, Analysis of assessment methods, Coordination on their territory of Community-wide quality assurance programmes organized by the Commission, Cooperation with other MS and the EC. Where relevant competent bodies shall comply with Section C of Annex I: QA/QC at national annd EU level, traceability, accreditation according to EN/ ISO 17025

7 AQUILA: role of NRL s EIONET Meeting Role and tasks of National Reference Laboratories Verifying and supporting the correct implementation of AQDs, by: Implementing a quality system in the laboratory Approving measurement systems (instruments, laboratories, networks) Ensuring the traceability of the measurements at national level, by providing/certifying reference materials to networks Organizing intercomparisons/round robin tests at national level Participating in EC QA/QC programmes Exchanging information through the organisation of training sessions, workshops, conferences and guidance documents AQUILA's role and the tasks of a NRL has been approved by DG ENV's "Air Quality Committee" in 2009 (download of document 'roles & requirements from ENV or AQUILA website).

8 AQUILA: structure EIONET Meeting Steering committee: chair, vice-chair and co-chairs Election of chair and vice-chair (4 years) Co-chair: DG ENV, JRC-IES (4 years) Secretariat: JRC-IES Chair Vice - chair Senior advisor U. Pfeffer LANUV NRW, DE F. Mathe EMD, FR P. Woods, NPL, UK Co - chair. DG ENV Co - chair A. Borowiak JRC

9 AQUILA: meetings EIONET Meeting st meeting: December th meeting: March 2011 focussing on, e.g.: - Accreditation of NRL s - Common PM equivalence tests - Development of CRM - Training on measurement uncertainty - PM2.5 measurement uncertainties

10 AQUILA: activities EIONET Meeting Examples of AQUILA s activities: PM QA/QC campaign ( ) VOC round robin test (2009) Co-Organisation of conferences and workshops JRC Intercomparison exercises in collaboration with WHO and AQUILA Production of documents/papers to topics of interest (e.g. guidance on equivalence) Contribution to implementation of AQ Directives (e.g. uncertainty of PM2.5 measurements to evaluate AEI)

11 11 Geostatistics in uncertainty evaluation of air quality data

12 Objectives of the study 12 Derive the uncertainty of spatially referenced measurements from their variogram analysis. 1. Develop a methodology for downloading geo-referenced air pollution data series of AirBase and for the automatic estimation of their variogram after discarding outliers; 2. Discuss the uncertainty of measurement data evaluated using the estimated nugget variance compared to the Data Quality Objective for a chosen pollutant (PM 10 ) 3. Identify trends over time in the nugget variance to show improvement (or worsening) of the uncertainty of measurement over the last ten years. 4. Create a warning system for outliers and questionable classification of monitoring stations

13 Variogram cloud 13 [ z x ) z( x )] 2 ( i+ l i

14 Lag Classes 14

15 Variogram 15 γ ( l) = 1 n( l) n l i+ 1 [ z( x ) z( x )] i+ l i 2

16 Variogram: spherical model 16 Range of spatially correlated measurements C 0 Nugget a

17 Nugget Variance 17 The nugget effect represents fluctuations of the measurements at very short distance (tending towards 0). s 2 meas is the variance associated with the sampling and analytical variability : the uncertainty of measurement s 2 sc 2 Nugget 2 Meas s = s + is the variance due to spatial variability that occurs at distances lower than the shortest sampling distance (microscale variance). Limit s 2 sc by determining s2 nugget of selected subsets of all sampling sites: background, rural in order to minimize s 2 sc s 2 sc

18 Outliers in a two (three?) dimension space michel.gerboles@jrc.ec.europa.eu 18 Neighborhood x, y n x = w 1 i x x i y i ±1 day ±2 º

19 Outlier analysis 19 Sx values (black), surrounding Mean (red line) and SD of all surroudings (blue line) sx = x x x i, y i x, y

20 Outlier test 20 Z + 2 Possible check of Classification Z - 2 z = x sx s sx z values of the station (black) plotted with the specified threshold (e.g. 2)

21 Uncertainty estimated from Nugget variance 21 Station/Station Area Type Nugget Values for the Year 2007 for PM 10 in µg/m³ Austria Germany France Italy Traffic urban Background-rural 1.9 Background-urban Background-Suburban Background stations ALL stations-station areas Comparison of standard Β gauge Field comparison of all Landers Field comparison of some networks? Data quality objectives (relative expanded uncertainty) PM % Limit/Target value Yearly: 40 µg/m³ Daily: 50 µg/m³ Corresponding uncertainty * 5 µg/m³ 6.3 µg/m³

22 Trend of nugget effect over 10 years 22 Station Combination Trend of the Nugget variance over 10 years ( ) Austria Germany France Italy Traffic urban -1.8% -0.2 % 0.6 % % Background-rural -0.7 % Background-urban -0.6 % -1.6 % -1.9 % Background-Suburban -0.3 % -1.8 % All Background stations -1.0 % -0.2 % -1.2 % -4,0%

23 Concluding remarks 23 Development of an automatic method for the indirect estimation of the uncertainty of measurements of the AirBase datasets. A tool for indicative testing of outliers and classification of station type has also been developed. First results indicate that uncertainties for PM 10 estimated using the Nugget variance are consistent with the DQO using all background stations. By including traffic type stations the uncertainties increased. Between , the trend of the nugget variance suggests a general decrease of uncertainty in measurement indicating an improvement of data quality and reporting data to AirBase. No trend on the spatial continuity (range) was observed. Seen the number of shortcomings of the method, its validation by comparison to a direct approach is needed. For now, this method can be used as a ranking tool.

24 Measurement Uncertainty in Air Pollution Data EIONET Meeting EUR (2010)

25 Source Apportionment Intercomparison EIONET Meeting European Intercomparison of Receptor Models for Source Apportionment of Air Pollutants European Reference Laboratory for Air Pollution

26 Source Apportionment Intercomparison EIONET Meeting Fulfillment of AQD requires identification of pollution sources Air Quality Assessment review every 5 years Application for postponement of attaining limit values (PM10, NO 2 ) Action Plans Quantification of: long range transport transboundary transport natural sources winter sanding and salting Source apportionment is relevant for the AQ Directive in a number of ways but the major aspect is related to planning. Before any effective mitigation strategy can be undertaken knowledge on the sources is required.

27 Source Apportionment Intercomparison EIONET Meeting Source Apportionment (SA) is the practice of deriving information about pollution sources and the amount they emit from ambient air pollution data. SA Methods: Basic numeric data treatment: correlation with meteorological parameters, subtraction of regional background Dispersion Models and Emission Inventories Receptor Models (based on statistical evaluation) Inverse modelling (multilinear regression) Back-trajectory analysis

28 Source Apportionment Intercomparison EIONET Meeting Survey of Receptor Models: Needs Study on comparability of different SA methodologies by inter-comparison of SA models Improve the expression of uncertainty (input data uncertainty, between model uncertainty, rotational ambiguity in factor analysis, source profile uncertainty, etc.) Evaluation and harmonisation of data collection and analytical schemes Development and Implementation of Quality Assurance and Validation protocols Creation of appropriate databases in terms of number of samples, time resolution and measured parameters Definition of common criteria for source labeling based on objective and quantitative information Improve identification of certain types of sources/ compounds: traffic (gasoline vs. diesel), ship emissions, biomass burning, biogenic secondary organic, nitrate component

29 Source Apportionment Intercomparison EIONET Meeting Step 0 JRC Inter-comparison for Receptor Models Open EoI and selection of participants (Summer 2010) Kick-off Workshop in (4 th -5 th November 2010 in Ispra) Step 1 Survey of receptor models suitable for the purposes of the inter-comparison Identification of the pollutants and metrics to test, according to the needs and to the most up-todate technical developments. Revision of the methodologies for uncertainty estimation and expression Definition of criteria for the assessment of model performance Discussion about the feasibility of a Common Protocol for source apportionment (including quality assurance procedures, validation criteria and quality standards) Step 2 Carry out an Inter-comparison between the involved research groups by applying the Harmonized Protocol and other widely accepted techniques to one or more common databases. Evaluate the outputs according to quality criteria and assess the influence of critical variables Check the influence of different scientific backgrounds/ approaches in source identification. Use the results of this exercise to set up common standards for the interpretation of receptor model outputs and to draft a common protocol to be used for obligations under AQD.

30 Source Apportionment Intercomparison EIONET Meeting Workshop programme: 4 th and 5 th November

31 FP7 project AirMonTech EIONET Meeting The new FP7 EU project AirMonTech compiles information to harmonize current air pollution monitoring techniques and to advise on future monitoring technologies and strategy. AirMonTech will gather information on instrument performance, test results, equivalence demonstrations and procedures for operation, maintenance and calibration, and process them into specifically designed databases. A roadmap for future urban air quality monitoring including recommendations on existing and new monitoring technologies will be developed and discussed with stakeholders.

32 FP7 project AirMonTech EIONET Meeting WP 1: Current technologies (EMPA) WP 2: Future technologies (IUTA) WP 3: Database (JRC) WP 4: Forward look (NPL) WP 5: Dissemination (ECN)

33 FP7 project AirMonTech EIONET Meeting AirMonTech Workshop December 2010, London Day 1: Measurement technologies for regulated compounds in air quality assessment Type approval, equivalence tests, QA/QC for PM monitoring, recommendations for EC/OC, NO2 & converters, sensors, laser-spectroscopic technologies Day 2: New metrics, technologies and strategies in air quality monitoring Health relevant metrics, ROS, PM surface area, bioaerosols, satellite remote sensing, personal exposure monitoring THANK YOU!