PM-LAB

0 downloads 0 Views 15MB Size Report
134. 3.6. Part 6: Mobile PM Measurements in Aachen, Maastricht and Liège . ...... Netherlands the Dutch NTA80 requires the cooling for the sampled filters. Presently ...... Bischopsingel and the traffic dosing light at the Willem Alexanderweg.
PM-LAB Towards a particulate matter information system for the Euregion

Index 1. Introduction ............................................................................................................................ 1 2. Monitoring aspects of PM-Lab ............................................................................................... 3 2.1 PMx Aspects ..................................................................................................................... 3 2.2. UFP Aspects ..................................................................................................................... 4 2.3. BC Aspects ....................................................................................................................... 5 2.4. On-line PM-Monitoring sites and choice of PM-Lab calibration sites ............................ 7 2.5. Reference Method Implementation ............................................................................. 12 2.5.1. Gravimetry: facilities, material and procedures .................................................... 12 2.5.1.2.Filter material and preconditonning .................................................................... 14 2.5.2. Reference samplers, related choices and issues ........................................................ 17 2.5.3. PM-LAB CALIBRATION ................................................................................................ 19 2.5.3.1. Experience design................................................................................................ 19 2.5.3.2. PM-Lab calibration: results, equations ............................................................... 21 2.6. UFP aspects of the project ............................................................................................ 23 2.6.1. Design and implementation of the UFP mobile-Lab .............................................. 23 2.6.2. Field monitoring using the UFP mobile-Lab ........................................................... 26 2.7. BC aspects of the project............................................................................................... 30 Appendix.......................................................................................................................... 33-95 3. Interreg IVa project PMlab Project report - Action 4 ........................................................... 96 3.1. Development of an interpolation model for the PM10 concentrations in the EMR .... 97 3.1.1. Preface .................................................................................................................... 97 3.1.2. Introduction............................................................................................................ 97 3.1.3. Development of an adjusted RIO model for the EMR ......................................... 100 3.1.4. Disaggregation of 5x5 km E-PRTR PM10-Emissions ............................................. 101 3.1.5. Calculation and optimization of Beta values .................................................... 102 3.5.2 Transmission term ................................................................................................. 102 3.1.5.3. Calculation of ‘Beta*²’ ....................................................................................... 102 3.1.5.4. Comparison between simple Land-Use Beta (RIO) and optimized Beta*² ....... 104 3.1.5.5 Mapping of daily PM10 levels ............................................................................ 105 3.1.5.6. Validation .......................................................................................................... 106 3.1.5.7. References ......................................................................................................... 108 3.2. Part 2: Web Mapping Tool .......................................................................................... 109 3.2.1 Introduction........................................................................................................... 109 3.2.2. Research motivation for the PM Lab project ....................................................... 109 3.2.3.Leitmotif for setting up and operating the web mapping platform ..................... 111 3.2.4. Exploration of the data flow................................................................................. 113 3.2.5. Implementation of the web-mapping platform ................................................... 115 ............................................................................................................................................ 115 3.2.6. Results and optimization opportunities ............................................................... 118 3.2.7. Conclusion ............................................................................................................ 121 3.2.8. References ............................................................................................................ 122 3.3. Part 3: Model comparison ........................................................................................... 123 3.3.1. Introduction.......................................................................................................... 123 3.3.2. Model comparison – preparation and technical background.......................... 123 3.4. Part 4: Additional Features complementing the web mapping system ..................... 129

3.4.1. Forecast tool for PM10 levels............................................................................... 129 3.4.2. Back-Trajectories .................................................................................................. 131 3.4.3. References ............................................................................................................ 131 3.5. Part 5: Additional Features complementing the web mapping system - Modeled Inversion Intensity .............................................................................................................. 132 3.5.1. Modeled Inversion Intensity – general information ............................................ 132 3.5.2. Typical evening cooling ........................................................................................ 133 3.5.3. References ............................................................................................................ 133 Annex.............................................................................................................................. 134 3.6. Part 6: Mobile PM Measurements in Aachen, Maastricht and Liège ......................... 136 3.6.1. Mobile Measurements in Aachen ........................................................................ 136 3.6.2. Mobile Measurements in Angleur ....................................................................... 136 3.6.3. Mobile Measurements in Maastricht................................................................... 138 3.6.4. Mobile Measurements in Liège / Centre ............................................................. 139 3.6.5. References ............................................................................................................ 139 3.7. Part 7: Data transformation ........................................................................................ 140 3.7.1. Data base .......................................................................................................... 140 3.7.2. SQL queries ........................................................................................................... 142 Appendix A – Data base ................................................................................................. 143 Appendix B – Scripts ....................................................................................................... 148 Appendix C – SQL queries............................................................................................... 150 3.8. Part 8: Clean Air Plans and emitters ............................................................................ 151 3.8.1. Introduction.......................................................................................................... 151 3.8.2. Interviews ............................................................................................................. 152 3.8.3. PM10 Emissions in the EMR ................................................................................. 154 3.8.3. Industrial PM10 Emissions in the EMR ................................................................. 156 3.8.4. PM10 Concentrations and Air Quality Planning in the EMR ................................ 166 3.8.4.2. Case Studies - Air Quality Planning in the EMR ................................................. 167 3.8.5. Summary and Conclusion Action 4.2 ................................................................... 175 4. Speciation – Source recognition – health effects............................................................... 176 4.1. Introduction ................................................................................................................. 176 4.2. Experimental approach ............................................................................................... 177 4.2.1. Chromium (Cr) ...................................................................................................... 177 4.2.2. Nickel (Ni) ............................................................................................................. 177 4.2.3. Platinum (Pt)......................................................................................................... 177 4.3. Results ......................................................................................................................... 178 4.3.1. Cr speciation ......................................................................................................... 178 4.3.2. Ni analyses ............................................................................................................ 181 4.3.3. Pt analyses ............................................................................................................ 185 4.4. General conclusions .................................................................................................... 186 4.4.1. Chromium ............................................................................................................. 186 4.4.2. Nickel .................................................................................................................... 186 4.4.3. Platinum ............................................................................................................... 186 4.5. References ................................................................................................................... 187 APPENDIX ....................................................................................................................... 188 5. Summary............................................................................................................................. 222 5.1. Consistent data............................................................................................................ 222 5.2. Health effects .............................................................................................................. 223

5.3. UFP measurements ..................................................................................................... 223 5.3. Heavy metals in particulate matter............................................................................. 225 5.3.1. Chromium (Cr) ...................................................................................................... 225 5.3.2. Nickel (Ni) ............................................................................................................. 227 5.3.3. Platinum (Pt)......................................................................................................... 228 5.4. Modelling – PM maps and dissemination of results ................................................... 229 5.4.1. Modelling.............................................................................................................. 231 5.4.2. The PMLab model for the EMR ............................................................................ 231 5.4.3. Data extraction and transformation .................................................................... 231 5.4.4. Mapping - PM10 maps on a daily basis ................................................................. 231 5.4.5. Validation ............................................................................................................. 231 5.4.6. Mapping ............................................................................................................... 232 5.5. Competent authorities ................................................................................................ 232 5.5.1. Competent authorities ......................................................................................... 232 5.5.2. Belgium ................................................................................................................. 234 5.5.3. The Netherlands – Province of Limburg ............................................................... 235 5.5.4. Germany - North Rhine-Westphalia (NRW) ......................................................... 236 5.6. Air quality plans ........................................................................................................... 237 5.6.1. Belgium - Federal level ......................................................................................... 238 5.6.2. Netherlands – Province of Limburg...................................................................... 239 5.6.3. Germany – North Rhine-Westphalia .................................................................... 239 5.6.4. Effectiveness of local air quality plans ................................................................. 240 5.7. Air quality monitoring ................................................................................................. 240 5.7.1. Belgium ................................................................................................................. 240 5.7.2. The Netherlands – Province of Limburg ............................................................... 240 5.7.3. North Rhine-Westphalia....................................................................................... 241 5.8. Conclusions.................................................................................................................. 242

1. Introduction Air pollution does not respect national borders. In the Euregion Meuse-Rhine (EMR), comprising parts of Wallonia, Flanders, the Netherlands and Germany, a three years cross-border project was executed on particulate matter (PM). In this Euregion area, the borders were relatively recently established, the national capitals are at distance and there is more mutual orientation. In recent years European unification has contributed to strengthen the Euroregion cooperation and the European Commission supports cross-border initiatives. The project was initiated by the Institut Scientifique de Service Public (ISSeP) Liège. It is well known that air quality measurements differ between measurement bodies, which leads to different measured levels for particulate matter, once a border is crossed. One complicating factor is that there exists no absolute calibration for PM. Not only measuring methods and values differ, even the interpretation of and dealing with the results in models and in situations with increased concentrations vary. Ultimately, this affects the assessment of exposure to particulate matter and air quality policies. In economic terms, this leads to an unequal playing field to develop activities where air quality limits may be exceeded, or where due to the high concentrations measures should be taken like a speed limit, or low emission zones. Project partners next to ISSeP were the Center for Environmental Studies (Applied and Analytical Chemistry, and Environmental Biology) from the University of Hasselt (CMK), the Geographisches Institute of RWTH Aachen University (Physical Geography and Climatology Group and Division for Economic Geography) and the Bureau Advice & Research (HMAO) of the Province of Limburg (PL), main partner in this project. The coordination was entrusted to CSE, a consortium of Zuyd University at Heerlen, GGD Limburg (Public Health Service) and HMAO (Province of Limburg). External partners are the VMM (Vlaamse Milieumaatschappij), LANUV Essen and the GGD Amsterdam. The project proposal was submitted in 2009 and approved. The project was made possible by grants from the European Regional Development Fund (ERDF), grants from ministries and by own resources of the partners. The objective of the PMLab project was to build an information system based on:  Establish consistent PM measurement results across borders  Research into the relevant sources and health effects  Mapping of PM exposure in the Euroregion by a web-platform  Dissemination of results to relevant authorities, enterprises and residents of the EM PM load exceeds limits

Health effects, Economic restrictions

Translation to air quality plans

Different methods in measuring, interpretation, different measures

More knowledge needed of PM sources, health effects, air quality plans

Advice on reduction measures, exposure

PM charts, insight in PM load, PR, internet

Speciation, effects, sources

Develop uniform methods: PMLab, comparison tests

Figure 1. Schematic outline of the PMLab project

1

In its design, the project is an example of how a regional environmental/social issue traverses a policy cycle starting with a problem definition, through to monitoring, presentation and policy advice. The project aims to provide information on the concentrations of PM and ultrafine particles, on the performance of state-of-the-art apparatus, and not to mention to foster cooperation between the different partners in the three countries, each with its own expertise and (work)-culture. In the first section of the summary the harmonization of the PM measurements is treated, followed by a short paragraph on health effects of PM. The third section gives an impression of the UFP measurements in the EMR. Section four adds knowledge to levels and health effects of some heavy metals in PM. The fifth section deals with the PM emissions, their sources, and a description of the modelling of PM levels, eventually resulting in publicly accessible online PM maps. Section six focuses on the (search for) competences of air quality authorities, in order to provide the most appropriate local or regional authority with relevant information. Section seven illustrates the use of air quality plans, the adequate instrument to reduce exposure to PM. Section eight gives an overview of the air quality measurement networks, and in section nine general conclusions are drawn. The full report follows the original division of actions by the different participants: Chapter 2 described the monitoring of PM and the UFP measuring campaign. Additional information on Carbon measurements can also be found in this chapter. Chapter 3 described the modelling of the consistent data, the web mapping tool and information on the different air quality plans in the EMR. In Chapter 4 information can be found on the speciation of Cr and Ni as well as the results from Pt analysis. The overall summary can be found in chapter 5.

2

2. Monitoring aspects of PM-Lab These tasks are Action 2 and Action 3. The partners carrying these actions (ISSeP and Provincie Limburg) are running ambient air Monitoring Networks. Action 2 is actually a preliminary part of action 3. Action 2 consisted in deciding upon, obtaining and launching the Ultra Fine Particulate mobile Lab. Action 3 entails two parts PM x (PM10 + PM2,5) Monitoring and UFP (Ultra Fine Particulate) Monitoring. During the prolongation period of the project (01/01/2013 to 30/06/2013) main effort has been put on BC (Black carbon) monitoring

2.1 PMx Aspects Legal reported data for compliance monitoring (checking observed PM 10 and PM2,5 concentrations against European Union Limit values, establishing the number of days exceeding the PM10 daily Limit Value is got:  From the results of the various on-line PM automatic analysers run by the various air quality monitoring networks across the Euregio  From runs of the standardised (EN12341(1998) promulgated for PM10 and the more demanding EN14907(2005) promulgated for PM2,5 gravimetric daily reference method, also called manual gravimetric method. Daily filters are collected with defined sampling characteristics, and daily concentrations are inferred from filter weighing results before and after sampling  More precisely, on site comparisons of on line analysers with the reference methods are performed. Calibration factors or equations are extracted from the comparison data, which adjusted the (validated) automatic analyser results to transform them into Equivalent, legally valid data. Discrepancies were observed between the various air monitoring networks of Euregio, coming from use of different on line automatic analysers, variations in the implementation of the gravimetric Reference Method (RM), differences in the design of the fixed air monitoring station networks (e.g. are they located in the most polluted areas or according to a grid?), and differences in the way to interpolate and get maps from point data. The aim of the PMx Monitoring work was to overcome these discrepancies to establish consistent calibration factors for all the various PM analysers present in the various networks of the Euregio. Thus, when applying these “PM-Lab” consistent calibration factors to on line PM analyser results, the resulting values should be consistent. The core of PMx Monitoring work has been:  To study, implement experiment and tune the best appropriate implementation of the reference gravimetric method. For this account is taken of traditions of the various networks of the Euregio, of the PM10 (EN12341-1998) and PM2,5 (EN14907-2005), of what is foreseen for the next merged PM10-PM2,5 updating standard.  To perform a long enough calibration campaign in selected fixed sites of the Euregio air network featuring a PM10 and or PM2,5 analyser.  By comparing results, to propose “PM-Lab calibration factors” for the various PM10 and PM2,5 analysers represented in the various networks. A comparison of results of PM-Lab’s implementation of the reference gravimetric method with that of the visited monitoring network results in a “PM-Lab calibration”. These calibration equations were delivered to Action 4 (Mapping) of Rwth-Geoklima. The PM-Lab Monitoring team was specifically assembled to be able to assist Action 4, with, among the ISSeP persons, a geographer and a modeller (the later also set up the PM-Lab database on behalf of Action4).

3

Complementary aspects  The PM-Lab Monitoring delivered inputs to the PM monitoring community. The CEN workgroup for the revision on the PM10 and PM2,5 standards (towards a merged standard) and the drafting of a technical specification for automated PM10 and PM2,5 analysers was active during the PM-Lab project, and ISSeP was represented in this workgroup.  The PM-Lab Monitoring team performed the sampling on behalf of Action 5 (Speciation) and assisted for the choice of sampling methods or characteristics

2.2. UFP Aspects Atmospheric aerosol particles cover a size range from few nanometres up to tenth of micrometers. While coarse particles larger than one micrometer are dominated by mineral dust, see salt and biological particles (like pollen, bacteria and spores) fine particles smaller than one micrometer are secondary material built from gases such as sulphates, nitrates and organic carbon or are originated from incomplete combustion processes such as elemental carbon (soot).

PMx

UFP

Figure 2. Relation between particle size and their impact on human health

Epidemiological and toxicological studies suggest that UFP may cause adverse human health effects greater than or independent of the effects compared to the larger particles. The size range of ultrafine particles smaller than 0.1 micrometer is dominated by soot particles, especially in urban areas. The concentration of such particles is highly variable and shows a significant daily and weekly variation, especially close to traffic. Soot particles often carry toxic trace compounds such as heavy metals and polycyclic aromatic hydrocarbons (PAHs). These compounds may erode in the lung and are then transported through the lining of the lung into the bloodstream. Although not all ultrafine particles contain such organic toxics, there is evidence that they can initiate oxidative stress in the lung. Oxidative stress is a process, which alters lung cell chemistry, causing inflammation and setting in motion a cascade of health problems. Ultrafine particle are also small enough to cross the lung membranes and reach the bloodstream. They can cause there immune responses such as thickening of the blood, which leads to an increased heart and stroke hazard. They deposit deeply in the lung and may cause cardiovascular and respiratory diseases. They can also be transported to different organs such as liver or heart via the blood stream. Such ultrafine particles are therefore of major concern to public health. Nevertheless, there is until now no regulations pertaining to their concentration in the ambient air. Existing measurements of UFP in urban areas in Europe are rare and are mostly driven by research institutes. There is not yet an International standard or reference method for measuring UFP.

4

The prescribed mass concentration limits for the particulate matter such as PM10 or PM2.5 are not relevant for ultrafine particles. Because of their small size they contribute only insignificantly to the PM10 or PM2.5 particle mass concentration. Considering the same density a 10 µm particle has indeed the same mass as one billion particles of 10 nm. Mass is thus not the good metric for UFP. Hence, the particle number concentration (PN) seems to be a better indicator to define the exposure to ultrafine particles in ambient air. There is thus a need for appropriate measuring instrumentation which is the basic reason of action 2 of the project. The action consisted in designing, acquiring, mounting and launching the Ultra Fine Particulate mobile Lab and then learning the best practice using this tool. Action 3 will be the deliverable of the project including first result of UFP levels in several locations within the Euregio, but also different collaboration and inter comparison with other network and scientific valorisation of the developed knowhow. The goals of the action UFP can thus be summarized as follow: - Training of the partners to acquire best practice in UFP measurement - Selection of the best available equipment - Collection of first result to have an indication of actual UFP levels - Initiate collaboration with scientific partners and other measuring networks - Setting up of a procedure for continuous monitoring of UFP - contribute to an International standard reference method - Comparison with other methods - Valorisation of the results through presentations at International conferences and papers

2.3. BC Aspects The influence of traffic on PM10 / PM2.5 is most of the time not predominant in urban environment. Indeed the majority of particles from vehicle exhaust are in the size range of 20-130 nm for diesel engines and 20-60 nm for gasoline vehicles (Morawska et al 1998, Ristovski et al 1998). These ultrafine particles (UFP), due to their very small size, do not contribute a lot to the total mass compared to coarse particles generated by other activities. Following PM is therefore not sufficient to show the impact of traffic on population or to highlight the effect of some reduction actions. As motor vehicle emissions usually constitute the most significant source of UFP (Hitchins et al 2000, Shi et al 1999). Most particles generated by the traffic are made of carbonaceous substances and result from an incomplete combustion of fuel. Different toxics like PAH or heavy metals are often linked to these soot particles. As this carbon substance has the property to absorb light it’s possible to correlate this absorption to the ambient level. BC is operationally defined as carbon measured by light absorption at a determined wavelength and could thus be used as an alternative metric of traffic pollution. BC presents the advantage to be quite easy to measure compares to UFP. Cheaper equipment with low maintenance and good time recovery are available.

© nasa.gov Figure 3. BC Emissions arising from industrial and Biomass Sources

The goals of the action BC include:

5

-

Selection of the best available equipment Measure the BC level at different type of location including traffic, urban, rural sites Prove the possibility to correlate traffic with BC measurement Find correlation between UFP, PM10 and BC

6

2.4. On-line PM-Monitoring sites and choice of PM-Lab calibration sites

fdms



Teom fdms

FH62

FH6 2 Bam

 FH62

 grimm180 opt Sharp



Figure 4.The PM monitoring sites selected by the mapping action and the variety of automatic analysers in use

There is some variation for the size selective inlets on automatic analysers. Generally the USEPA (1m³/h) are used. However these tend to be replaced in the Netherlands by European (1m³/h) inlets, calculated or downsized from the European standardized PM10 and PM2,5 (2,3m³/h) inlets. PM10/PM2,5 analysers in use in the various ambient air monitoring networks in the Euregio

7

Table 1. Overview of the different devices and measurement principles

Measurement principle Beta Attenuation

TEOM ®

Optical

Devices and manufacturers

Remarks

FH62IR from Eberline (DE) TE62IR from Thermo (USA) SHARP from Thermo (USA)

Earlier devices, Thermo absorbed Eberline. Very similar or identical devices Most up to date device of Thermo, Includes also a nephelometer for a short time resolution indicative result Recently introduced (in Euregio)

BAM Beta Attenuation Monitor of Met one (USA) MP101M.C from Environnement (FR) Teom or Teom-SES (Thermo) Teom-fdms (Thermo)

Grimm 180 from Grimm Aerosol (DE)

Limited to 24h cycles to be equivalent, to be abandoned The successive evolutions of the instruments aim at minimising loss of volatile fraction and underestimation compared with reference gravimetric method. Fdms is the latest of those evolutions See below

Table 2. Other relevant precisions

PM analysers flows and size selective inlets Other PM parameters

PM analysers generally operate at the US (or USEPA) standard flow of 1m³/h with a size selective inlet (that is either with a PM10 inlet or with a PM2,5 inlet). Thus two separate devices are necessary to monitor the two regulated size fractions, PM10 and PM2,5). The Grimm 180 operates at 72 liters per hour without any size selective inlet and delivers PM10, PM2,5 and PM1 concentrations with a single device. Black Smoke measurements (although not envisaged at all any longer by directive 2080/50 ruling in particular PM10 and PM2,5 assessment) are being continued at some sites because of their health effects relevance. In the Netherlands and in Germany, the Reference gravimetric method is run continuously or semi-continuously in particular to replace (PM2,5) analyser results which can be no longer validated, of appropriate PM analyser which have not yet been delivered.

Measurement principles of analysers in use in this larger Euregio area are Beta attenuation, Teom ® (Tapering Element Oscillating Monitor), and optical analysis represented by the Grimm 180 (PM10 and PM2,5 concentrations estimated from particulate number concentration for 31 size classes). Networks intentions are to be taken into account. For Lanuv-Nrw, the most likely choice for new (replacement) analyser order is the Sharp and in the Netherlands it is the BAM The Monitoring action must select for its PM-Lab calibration, a limited number of monitoring sites out of the total population of sites used by the Mapping action. These PM-Lab calibrations by the PM-Lab implementation of the reference method will consist in:  Measuring PM10 at 6 main calibration sites during close to one year  With representative PM2,5 and PM10 comparison campaigns, to get a PM-Lab estimate the relationship (equation) between network on-line analyser results and the gravimetric reference method results This must be done independently from the various air networks own work or efforts with the gravimetric reference methods, because of the variations, across networks, in the implementation of the gravimetric reference method. These can pertain to filter material, sampler used (or version of sampler with/without exposed filter cooling), filter handling or conditioning…)

8

Table 3. Automatic Monitoring stations used for Action 4 Mapping

Name

Location

AABU VACW

Aachen-Burtscheid Aachen-Wilhelmstr.

NIZI

Niederzier-Treibbach

EIFE BONN

Simmerath Eifel Bonn-Auerberg AnderJosefhoele

DDCS

Duesseldorf-Corneliusstr

LOER GRGG HUE2 VKCL RODE KREF LEV2 VMGR MGRH

Station type Network Urb bgrd Trafic Industri al Remote Urb bgrd

LANUVTrafic NRW Urb Duesseld-Loerick bgrd Industri Grevenbroich-Gustorf/Gindorf al Industri Huerth Dunantstr al Köln Clevischer Ring Trafic Urb Köln Rodenkirchen bgrd Urb Krefeld Linn bgrd Urb Leverkusen Manfort bgrd Moenchengldb Duesseldorferstr Trafic Urb Moenchengldb Rheydt bgrd Urb Nettetal Kaldenkirchen bgrd

NETT DERP01 5 Westeifel Wascheid DERP02 1 Neuwied Hafenstraße 42N016 Dessel 42N035 Aarschot

42N045 Hasselt-Schepvaartkaai 42N054 Landen-Walshoutem 40GK06 Diepenbeek 40GK09 Genk

Remote Industri RheinLan al d-Pfalz Industri al Rural bgrd Urb bgrd VMM Rural bgrd Urb / Indu Industri al

9

PM 10 analys.

PM 2,5 analy s.

SHARP SHARP Teom-ses Teomfdms Teomfdms Teomfdms Teomfdms

ref ref PM2, PM10 5

black smok e

0.5 0.5 1 SHAR P

X SHAR P

Teom-ses Teomfdms Teom-ses

0.5

0.5

1

0.5 1

0.5

FH62IR Teom-ses FH62IR Teom-ses

SHAR P

0.5

SHARP SHARP

TEI62IR TEI62IR TEI62IR TEI62IR Teom Teomfdms

TEI62I R

x

Urb Prov. Geleen Geleen Asterstraat bgrd BAM BAM Limburg PLIM2 Maas A2 Nassaulaan Trafic NL BAM BAM RIVM13 Rural 1 Vredepeel Vredeweg bgrd FHI62IR 1 1 x RIVM RIVM13 Rural 3 Wijnandsrade Opvergeltstraat bgrd FHI62IR RIVM13 Urb 7 Heerlen Deken Nicolaye straat bgrd FHI62IR 1 RIVM13 6 Heerlen Looierstraat Trafic FHI62IR 1 RIVM23 6 Eindhoven Genovevalaan Trafic FHI62IR RIVM23 7 Eindhoven Noordbrabantlaan Trafic FHI62IR RIVM23 Rural 0 Biest Houtakker Biestsestraat bgrd ? Industri 43R240 Engis al Grimm 180 x Urb / mp101m /Grimm 43R221 Herstal Liege Indu * Rural 43N063 Corroy-le-Grand bgrd Grimm 180 Rural 43N073 Vezin bgrd mp101m Rural ISSeP 43N093 Sinsin bgrd Grimm 180 Rural mp101m /Grimm 43N113 Sainte-Ode bgrd * Rural 43N066 Membach (replaces Eupen) bgrd Grimm 180 Rural 43N67 Membach bgrd Grimm 180 Urb / 43R223 Seraing Jemeppe Indu Grimm 180 Urb mp101m /Grimm 43H201 Saint-Nicolas bgrd * Urb 43R201 Liege-Congrès bgrd Grimm 180 Urb / mp101m /Grimm 43R222 Liege-issep Indu * x Urb / 43M204 Liege Angleur Indu Grimm 180 1 =every day 0.5 = every second day Teom-fdms: in DE and the NL, Teom-fdms PM2,5 values are not further validated/accepted. Reference method is used instead. (*) mp101m/Grimm: up to now an mp101m (PM10), arrival of a Grimm 180 expected for April 2011 AABU PM-Lab Calibration site

10

Aachen Burtscheid

Hasselt Bam inlets

Maastricht A2

Swam inlets SEQ Leckel

Heerlen Angleur

Grimm cabinet

SEQLeckel

SEQLeckels

Membach

Figure 5. The six PM-Lab calibration stations and the PM-Lab Reference samplers

Height of inlets of the Reference samplers must be as possible similar to that of the automatic analysers. Thus use of an elevating stand (Aachen), installation of samplers on the roof of the automatic cabin (Hasselt) or use of an extended sampling line (Maastricht). This was not done at all sites (see Angleur).

11

2.5. Reference Method Implementation 2.5.1. Gravimetry: facilities, material and procedures 2.5.1.1. Weighing facilities When starting in the fall of 2009: - At ISSeP, there was an existing facility (glove box and balance in a temperature conditioned balance room), but its capacity of 40 filters per batch, was entirely used by the calibration for the Walloon network by the gravimetric reference method, so that was no capacity left for PM-Lab - Provincie-Limburg did not have any weighing facility appropriate for ambient air filters. At ISSeP The air networks, which needed an extra weighing capacity anyways, paid the new equipment, the PM-Lab persons took care of everything. Mettler won the tender end 2009 and the Mettler XP-26 balance was delivered in February 2010. We went for the solution already used for the existing balance: thus a second glove box in the existing temperature conditioned balance room, with control of the RH in the glove box by salts baths. Accessories were designed and ordered: large capacity, custom made towers: each tower contains up two 72 filters, placed on six trays containing 12 filters each. Six towers were ordered. Each filter place on any tray of any tower is identified. Each filter occupies the same place when weighed as virgin filter and after sampling as exposed filter. The glove box contains 2 towers at the same time. The actual limit for batches is 144 filters.

Figure 6. Upper Left; New glove box at ISSeP, Upper right; Operator at the balance (ISSeP) Lower left; New balance at ISSeP, Lower right; Filters tower (ISSeP)

Relevant data of appropriate sensors (room temperature, glove box temperature +RH) is logged on a server and permanently accessible to staff members in charge (it’s called the supervision system).

12

1

2

3

Figure 7. Example of screen of the ISSeP supervision system

On this screen, the influence of various activities on glove box conditions can be seen. The period displayed correspond to the conditioning and to weighing of virgin filter for “Batch n°8” of trials (1923/11/2010). Blue line = Temperature (°C) in the glove bow – Green line = RH in the glove box. Moment 1 is the introduction of the two towers with virgin filters in the glove box. Moment 2 is the first weighing for these virgin filters Moment 3 is the second weighing for these virgin filters. Tuning of this equipment involved mainly: Tuning of the salt solution mix. RH actually measured does not correspond to what the theory says about the RH directly above the saturated salt solution; Installation of fans pushing up in the glove box the RH adjusted air from immediately above the salt baths; Control of RH at various heights in the glove box. The range in standards presently in force is 20°C +/- 1°C and RH of 50% +/- 5%. In the draft for the new merged standard the RH range became 47,5% +/-2,5%. This was taken into account when adjusting the glove box conditions: the choice was to tune directly the new glove box in the range of the foreseen merged standard. At Provincie Limburg, Maastricht facility. The weighing system was installed in the building of the Province in Maastricht (the building in Heerlen were the Monitoring services are installed was found to unfavourable because of vibrations).

Figure 8. LEFT; Filter shelves in the Maastricht filter conditioning room (20°C, 50%RH), RIGHT; Enclosure in that room for the 100% RH pre-conditioning prescribed by NTA 8019)

13

2.5.1.2.Filter material and preconditonning With gravimetric standards presently in force, use of quartz is mandatory for PM10 while other filter materials are also authorised for PM2,5: fibre glass, PTFE or a combination of PTFE and fibreglass. In the Netherlands, a uniform implementation of the gravimetric reference method was laid in NTA 8019 (Nederlands Technische Afspraak, January 2008) which requests the Whatman quartzfibre (QMA) to be used. From trials performed for the updating of the gravimetric standards, in particular by Vmm, it appears that compared with other quartz brands, use of Whatman QMA results in higher values. This is attributed to other materials used as binder (borosilicate) in the Whatman QMA, which stimulates retention of (semi-)volatiles on the filter, and that is considered as a positive artefact. Vmm decided to use the Pall Tissuquartz (Pall reference), which is considered as the most pure (binder free) and artefact free. In its report “Comparative PM10 and PM2,5 measurement in Flanders, period 2008-2009) Vmm estimates that for PM2,5 use of the Pall Tissuquartz results in values inferior by 6% to those got with Quartz Whatman. In 2010, GGD warned the monitoring community about considerable differences resulting from the use of specific batches of the Quartz Whatman QMA. PM Lab decided to choose the Pall Tissuquartz in the spring of 2010. In the fall of 2010, the Walloon Ministry asked ISSeP to use also the Pall Tissuquartz for the (now mandatory) yearly revision of its analyser calibration (which had been previously performed with Quartz Whatman). In the summer of 2011, a 15 days (July 21-August 4) experience PM10 in front of ISSeP comparing the quartz filters Whatman QUA and Pall Tissuquartz showed a Tissuquartz/QMA ratio of the means 0,86 (17,1 µg/m³ versus 19,8µg/m³), with a clearly systematic overestimation of the QMA compared to Tissuquartz. Quartz Whatm an vs Pall Tissupartz 30

Pm10 µg/m³

25 20 15 10

Whatman Pall

5 0 21/07

23/07

25/07

27/07

29/07

31/07

Figure 9. Comparison Pall Tissuquartz-Whatman QMA (July-August 2011)

14

2/08

4/08

Table 4.PM-Lab implementation of the Ref Method with reference to the draft merged PM10-PM2,5 standard

Topic Field blank value

Field blank place Certif. Ref. masses Resident filters

%RH, glove box sensors 3rd weighing Preconditioning at 100% RH

Visual check of virgin filters (elimination of filters with flaws) Blowing away possibly loose fibres of virgin filters Cleaning of rings for fibres before reuse

Static discharger

Draft update for standard pr EN12341 Should be < 60µg but no invalidation Results must not be corrected Loader or unloader

PM-Lab decision

“Weighing room blanks”, minimum 2 of them Mandatory – Deviations should be < 40 µg Range becomes to 45%-50%

Ok

Virgin mandatory if diff. ≥40µg Exposed mandatory if diff. ≥60µg According to outcome of suitability tests for filters) should be done. Duration: 3 weeks

Mandatory

Ok

Unloader Yes, issep:100 µg ;PL: 200 µg

Issep glove box already in that range PL weighing room can be adjusted Ok PL: done but limited to 14 days ISSeP: done in an exsicator (where filters remain in piles), done using with fully appropriate filter towers only from 2012 Done (both PL and ISSeP)

Found useful for Pall Tissuquartz filters (no binder!) but actually implemented only by PL

Filter dependent mandatory

The leaner Swam filter cartridges (without any joint) retain sometimes fibres afterwards These filter cartridges were swept for fibres before reuse Implemented (imbedded with doors opening of the Mettler balance)

Practical constraints for preconditioning were the following: for PL the limited place in the 100%RH enclosure and at ISSeP the two extra “filter towers” for this were ordered early in 2011, but delivery suffered of very large unexpected delays.

15

Figure 10. LEFT; Exsicator for 100% RH, RIGTH; preconditioned Box (lid removed for picture) for 100% RH preconditioning of a tower of virgin filters. Distilled water at the bottom, tower set on a register above the water surface

16

2.5.2. Reference samplers, related choices and issues A decision had to be made regarding the reference samplers to use for the calibration campaigns. Preparatory experiences (July 2010-jan 2011) oriented that choice. Complementary experiences (Feb-July 2012) studied unresolved questions regarding samplers. Here below only the Leckel sequential sampler actually used to obtain the PM-Lab analyser calibration equation is presented. In Belgium, the non cooled version in used, whereas in the Netherlands the Dutch NTA80 requires the cooling for the sampled filters. Presently that cooling is mandatory only for PM2,5, but the proposed merged standards extends that obligation to PM10. Leckel’s factory settings: cooling starts when temperature in jacket around filter cylinder measured is above 20°C and stops when it is lower than 18°C Cooled Leckel

Non cooled Leckel (door opened) See cylinders for virgin (left) and sampled (right) filters

Cooled jacket (left)around sampled filter cylinder

Airco element

Angleur (preliminary trials 2010) – two inlets of the Cabin Swam with one inlet of the Cabin SEQLeckel

Leckel PM10 inlet + sheath air tubing

Swam PM10 +PM2,5 inlets, Sampling line Insulation (in lower part)

Figure 11. Overview of the different sampling equipment used in the PMLab project.

17

In Germany, samplers are often placed inside the cabin of gaseous analysers, were they benefit of the cabin’s general airco. That implementation was tried in Angleur both for non-cooled and for the alternative Swam 5aDouble Channel analyser and sampler

18

2.5.3. PM-LAB CALIBRATION 2.5.3.1. Experience design Plans regarding this evolved. The first idea in 2009 regarding for that Euregio wide calibration was to run a PM-Lab trailer, with much embarked equipment, for extended periods, collocated at chosen representative sites with PM measurement of networks at this station. This would allow more extensive comparisons than the 14 days ones done across Europe carried by the ERLAP (European Reference Laboratory for Air Pollution, Environment Institute, Ispra Joint Research Centre) were of 14 days duration. In the fall of 2009, however, a discussion where the VMM and the GGD Amsterdam air monitoring networks were represented produced important changes in the PM-Lab approach. The choice then made was to set up a long comparison of on-line analysers run the various air networks with the reference method as optimised by PM-Lab, at a number of representative sites. The role of the Mobile Laboratory (in a trailer) was evolved towards that of an Ultra Fine Particulate mobile laboratory. Preparatory experiences, explored the effect of variations of the reference methods, before PM-Lab method details were set up. For this, finalised weighing facilities were necessary (which were ready in May 2010 at ISSeP and September 2010 in Maastricht). From the time equipment was operational, a few weeks were needed, for both organisations, to assess own performance for the gravimetric method, also to trouble shoot and streamline this work. Validated data of these preparatory experiences begins around April 2010, at the Angleur site and in October at the Maastricht-A2 site. Four Leckel sequential samplers (without the exposed filter cooling), ordered for the project, were received in February 2010. These samplers were to begin with used for tests regarding filter type, flow choice and sampling duration for samples to be collected for the speciation action.The two Swam DC samplers and beta attenuation analysers ordered for the project were installed and launched in the air networks cabin of Angleur and Maastricht-A2 in May and June 2010. A first selection delivered the following list: Table 5. Choice regarding sites, samplers and PM fraction site Site type PM10 analys. PM 2,5 analys. Region NRW LANUV network AABU Aachenurban Sharp Burtscheid background EIFE Remote (Teom-Fdms) Sharp Simmerath Eifel Region Limburg NL Maastricht-A2 trafic Bam Bam Plim2, Provincie Heerlen Deken urban FH62 IR Rivm 137 background Region Flanders Vmm 42N045 urban FH62IR FH62IR Hasselt background Region Wallonia ISSeP 43M204Angleur urban Grimm 180 industrial 43n67 rural Grimm 180 Membach background 1 = reference gravimetric method every day 0,5 = reference gravimetric method every other day

netw. grav PM10

netw. Grav PM2,5

1 0,5

0,5

1

Choices had to be made regarding the PM fraction chosen (PM10 or PM2,5) for our PM-lab comparison or calibration. Limit values are more clearly defined for PM10 than for PM2,5. On the other end, since PM10 is made of PM2,5 and an added coarse contribution (PM10-PM2,5) which is more local. PM2,5 is intrinsically more favourable for mapping, and is more relevant for heath effects.

19

Lanuv and Rivm do not validate any more their Teom-fdms results, the data loss is mitigated by use of the reference method at some sites. Thus the site of Simmerath was dropped, since only PM2,5 analyser (Sharp) analyser was still in use and low values (unfavourable for a calibration) were expected. For the Grimm 180 which had been recently installed in Eupen, re-suspension from vehicle on the non-covered road (or path) along the station caused high peaks and the results could not be validated. AwAC and ISSeP decided to install that Grimm in an own cabinet at a new site, Membach, rue du Moulin 4; sewage treatment station. The network Grimm data there began on February 15, 2011. Samplers available were the four ordered Leckels (without the exposed filter cooling), up to four stand-alone Swams and two ordered Swam (to be installed in network cabins). Comparisons for the PM-Lab calibration began, regarding PM-Lab’s reference instruments, with a combination of non cooled Leckels and Swams. Within a few months however, the two Swams in own cabinet at Membach (n°139) and at AachenBurtscheid (n°137) broke down. Another concern was the Swam versus Leckel differences observed in 2010, which were more marked for Swams in network cabins than for stand alone Swams in own cooled cabinet. Swams at Membach and Aachen-Burtscheid were replaced (for PM10) by Leckel PM10 samplers. At the Maastricht A2 station, a Leckel in parallel with the Swam in the network cabin had been foreseen anyways and at the Angleur site the calibration of the Grimm analyser done by the ISSeP network delivered much PM10 and PM2,5 data. Accordingly, only the (non cooled) Leckel results are finally taken into account for the PM-Lab calibration. Issues with the Swam are presented and discussed in an appendix. Table 6. Final comparison Set-up for the PM-Lab calibration Site + type Netw. analysers Leckel of PM-Lab AABU Aachen-Burtscheid urban background

Leckel installation details

Sharp PM10

PM10 Leckel

raised

Maastricht-A2 Plim2, Provincie trafic Heerlen Deken Rivm 137 Urban backgrond

Bam PM10

PM10 Leckel

Bam PM2,5

PM2,5 Leckel

Extended (sheath) line

FH62IR PM10

PM10 Leckel

42N045 Hasselt Schepvaartkaai urban background Region TMLG04 Angleur urban industrial TMNT10 Membach rural background

TE62IR PM10

PM10 Leckel

TE62IR PM2,5

PM2,5 Leckel

Wallonia Grimm PM10 180 PM2,5 Grimm 180 PM10

ISSeP

Roof of network cabin

Ground Level PM10 Leckel

20

2.5.3.2. PM-Lab calibration: results, equations The mean for the RM gives an idea of concentrations, and the ratio of the means RM/Cand is the most obvious or immediate correction factor. These are means for the PM-Lab comparison days. Table 7. PM10 calibration equations

µg/m³ mean Ref mean cand Sharp PM10 17.0 302 Aachen-Burtscheid 17.1 BAM PM10 24.4 220 Maastricht-A2 30.8 FH62IR PM10 16.2 198 Heerlen Rivm137 13.0 FH62IR PM10 22.9 293 Hasselt N045 17.7 FH62IR PM10 20.2 491 Heerlen + Hasselt 15.8 Grimm PM10 28/01/2011 29.0 Angleur 43M204 19/07/2012 32.3 Grimm PM10 16/04/2011 15.0 237 Membach 43N067 26/01/2012 18.8 Grimm PM10 official issep calibration equation 2011 Start End 2/04/2011 22/02/2012 2/05/2011 15/05/2012 16/04/2011 13/01/2012 8/02/2011 26/01/2011

common days

ratio RM/ Cand

PM-Lab calibration equation

0.99

Yc = Yu

0.79

Yc = 0,903Yu - 3,36

1.25

Yc = 1,21 Yu

1.29

Yc = 1,26Yu

1.28

Yc = 1,27Yu

0.90 0.80

Yc = 0,70Yu +2.0 Yc = Yu – 2,7

For the Heerlen FH62IR PM10 data, the last comparison period (Jan 28 to February 22, 2012) not taken into account (dumped from data set, details under 2 b4). The PM-Lab calibration equation of the FH62IR PM10 is that of the merged data set above (Heerlen + Hasselt). The VMM is designated as TE62IR or FH62IR. We do not know the meaning of these different names. Grimm PM10 The official issep equation for 2011 was got from 4 sites including Angleur (same implementation of the reference method as Pm-Lab’s). The Pm-Lab Angleur equation is close to that 2011 issep equation, whereas PM-Lab’s Membach equation is very different. The PM-Lab calibration equation chosen is the official issep 2011 equation for all sites, but for Membach (own Pm-Lab calibration equation retained). Table 8. PM2,5 Calibration equations

Start

BAM PM2,5 Maastricht A2 FH62IR PM2,5 Hasselt N045 Grimm PM2,5 Angleur M204

End 17/02/2011 1/04/2011 8/02/2011 26/01/2011 28/01/2011 19/07/2011

µg/m³ common mean Ref

PM-Lab

days

calibration equation

ratio Ref / mean cand Cand 32.3 0.92 35.2 14.5 1.03 14.1 20.0 0.79 25.2

35 269 201

Grimm PM2,5 official issep calibration equation 2011

21

Yc = 0,91Yu Yc = 1,18Yu - 2,1 Yc = 0,91 Yu - 2,9 Yc = 0,97 Yu - 4,12

Grimm PM2,5 The official issep equation for 2011 was got from 4 sites including Angleur (same implementation of the reference method as Pm-Lab’s). The Pm-Lab Angleur equation is close to that 2011 issep equation, whereas PM-Lab’s Membach equation is very different. The PM-Lab calibration equation chosen is the official issep 2011 equation for all sites, but for Membach (own Pm-Lab calibration equation retained).

22

2.6. UFP aspects of the project 2.6.1. Design and implementation of the UFP mobile-Lab The prescribed mass concentration limits for the particulate matter such as PM10 or PM2.5 are not suitable for ultrafine particles. The particle number concentration (PN) seems to be a better indicator to define the exposure to ultrafine particles in ambient air. This was the basic reason of action 2 of the project. Existing measurements of UFP in urban areas in Europe are rare and are mostly driven by research institutes. There is also no International standard or reference method for measuring UFP. The choice of devices used in the UFP mobile-Lab is the result of a long selection process based on thorough literature investigation, contact with manufacturers and users (monitoring networks and research institutes) and comparative tests. - Comparison of available methods and devices: After a first selection based on the literature, comparisons of various instruments (TSI SMPS, Grimm SMPS, Palas SMPS, TSI 3031, ELPI+ (Dekati), Phillips Nanomonitor) have been performed. Results obtained for the different techniques are showed below (details: presentations 1,, 3 and 5 - UFP appendix).

ELPI +

TSI 3031

S M PS

CPC

20000

AE 22

TNC

Phillips Aerosense 18000

DMA

16000

TSI3031

14000

ELPI1

12000

SMPS

10000 8000

TSI 3031

6000 4000 2000 Time 0

Figure 12. Comparison of candidate UFP devices

23

Table 9. Pearson correlation between instruments obtained during comparison tests

12,5%. It does not deliver equivalent data. 4

iv) Last we have to check whether making the intercept correction on top of the slope correction improves the agreement between corrected an analyser results and reference results. We write Yc = 0,90Yu – 3,36 (Which is deduced from Yu = 1,11Yc + 3,72 or Raw analyser results = 1,11 Ref.Method.Result +3,72) And we see (lower right quarter of table) that the resulting uncertainty for the corrected data is improved. It becomes 6,43%50 µg/m³ (Table 4, below).

D0 forecast : modeled class values – measured class values model class deviation mean PM10 = 2

= 2

= 2

34% 0% 51% 6% 33% 4% 44% 0% 17% 0% 19% 0% 26% 2% 33% 0% 34%

D2 forecast : modeled class values – measured class values model class deviation mean PM10 = 2

= 2

= 2

23% 4% 45% 9% 37% 4% 48% 2% 21% 2% 15% 4% 32% 2% 33% 10% 36%

LE

6

8

34

34

3

1

5

5

D0 forecast : modeled class values – measured class values

station AABU EIFE NETT 43N085 43R201 43M204 42N045 RIVM133

Aachen-Burtscheid Simmerath Eifel Kaldenkirchen Vielsalm Liège-Congrès Angleur Hasselt Wijnandsrade Average

CHIMERE D0 = 2

22% 0% 44% 44% 33% 0% 56% 0% 22% 0% 0% 11% 11% 11% 22% 11% 36%

= 2

0% 0% 56% 33% 0% 11% 11% 11% 22% 0% 11% 0% 0% 0% 11% 0% 21%

= 2

0% 0% 33% 11% 33% 0% 11% 0% 0% 0% 11% 0% 0% 11% 11% 0% 15%

D2 forecast : modeled class values – measured class values model class deviation station AABU EIFE NETT 43N085 43R201 43M204 42N045 RIVM133

Aachen-Burtscheid Simmerath Eifel Kaldenkirchen Vielsalm Liège-Congrès Angleur Hasselt Wijnandsrade Average

CHIMERE D2 = 2

22% 0% 67% 33% 22% 0% 67% 0% 11% 0% 11% 0% 11% 11% 22% 11% 36%

= 2

0% 0% 56% 22% 11% 0% 11% 0% 11% 0% 0% 0% 0% 0% 0% 0% 14%

= 2

0% 0% 56% 0% 33% 0% 33% 0% 0% 0% 0% 0% 11% 0% 11% 0% 18%

HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model Access via NOAA ARL READY Website (http://ready.arl.noaa.gov/HYSPLIT.php )

Summary: D0 + D2 Forecast for days with ≥ 3 stations >50 µg/m³

model class deviation

HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model Access via NOAA ARL READY Website (http://ready.arl.noaa.gov/HYSPLIT.php )

Summary: D0 + D2 Forecast for all 48 comparison days

Table 5+6. Summary of model deviations f0r D0 and D2 forecasts for weather type 1 (Table 5, top) and weather type 15 (Table 6, below)

D0 forecast : modeled class values – measured class values model class deviation station AABU EIFE NETT 43N085 43R201 43M204 42N045 RIVM133

Aachen-Burtscheid Simmerath Eifel Kaldenkirchen Vielsalm Liège-Congrès Angleur Hasselt Wijnandsrade Average

CHIMERE D0 = 2

36% 7% 36% 57% 57% 14% 50% 29% 57% 7% 36% 14% 36% 14% 57% 7% 64%

= 2

21% 14% 50% 36% 29% 21% 21% 7% 21% 7% 14% 7% 14% 7% 21% 7% 38%

= 2

29% 0% 43% 14% 21% 14% 21% 0% 21% 0% 21% 0% 29% 0% 14% 0% 29%

D2 forecast : modeled class values – measured class values model class deviation station AABU EIFE NETT 43N085 43R201 43M204 42N045 RIVM133

Aachen-Burtscheid Simmerath Eifel Kaldenkirchen Vielsalm Liège-Congrès Angleur Hasselt Wijnandsrade Average

CHIMERE D2 = 2

29% 7% 57% 36% 43% 7% 71% 7% 50% 0% 29% 7% 29% 14% 43% 7% 54%

= 2

14% 14% 43% 21% 7% 14% 7% 7% 21% 7% 21% 7% 7% 7% 7% 7% 27%

= 2

7% 7% 43% 14% 21% 14% 36% 0% 14% 7% 7% 14% 14% 7% 21% 7% 29%

D0 forecast : modeled class values – measured class values

station AABU EIFE NETT 62N085 62R201 62M204 42N045 RIVM133

Aachen-Burtscheid Simmerath Eifel Kaldenkirchen Vielsalm Liège-Congrès Angleur Hasselt Wijnandsrade Average

CHIMERE D0 = 2

41% 14% 55% 23% 52% 14% 35% 35% 48% 22% 52% 22% 64% 18% 52% 26% 71%

= 2

= 2

50% 0% 64% 5% 43% 0% 70% 0% 22% 0% 22% 0% 32% 0% 39% 0% 43%

D2 forecast : modeled class values – measured class values model class deviation station AABU EIFE NETT 61N085 61R201 61M204 42N045 RIVM133

Aachen-Burtscheid Simmerath Eifel Kaldenkirchen Vielsalm Liège-Congrès Angleur Hasselt Wijnandsrade Average

CHIMERE D2 = 2

18% 5% 45% 0% 33% 5% 35% 0% 26% 0% 30% 0% 41% 0% 43% 0% 35%

= 2

32% 9% 36% 14% 33% 5% 48% 4% 22% 9% 22% 9% 23% 9% 13% 13% 37%

= 2

36% 5% 55% 5% 43% 0% 52% 4% 22% 0% 17% 0% 36% 0% 26% 13% 39%

HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model Access via NOAA ARL READY Website (http://ready.arl.noaa.gov/HYSPLIT.php )

Summary: D0 + D2 Forecast for days with weather type 15 (23 days)

model class deviation

HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model Access via NOAA ARL READY Website (http://ready.arl.noaa.gov/HYSPLIT.php )

Summary: D0 + D2 Forecast for days with weather type 1 (14 days)

Examples of modeled PM10 concentrations in the EMR are shown in Fig. 2 and 3 for a day with easterly wind directions and elevated PM10 levels. Model outputs are complemented by meteorology and back trajectories showing the origin of advected air masses. Figure 2+3. Model outputs for D0 forecast (Fig. 2, top) and D2 forecast (Fig. 3, below)

Wijnandsrade: 57 Hasselt: 54 Liège: 49 Vielsalm: 33 29.01.2011 D0 forecast

weekday Sat

HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model Access via NOAA ARL READY Website (http://ready.arl.noaa.gov/HYSPLIT.php )

PM10 (µg/m³)

DWD AABU (Aachen) 205 m EIFE (Simmerath) 572 m date Weather type PP (hPa) TT (°C) UU (%) RR (mm) FF (m/s) DD (°) PP (hPa) TT (°C) UU (%) RR (mm) FF (m/s) DD (°) 29.01.2011 1 994 -2,8 85,1 0,1 1,5 58

22

Wijnandsrade: 57 Hasselt: 54 Liège: 49 Vielsalm: 33 29.01.2011 D2 forecast

weekday Sat

HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model Access via NOAA ARL READY Website (http://ready.arl.noaa.gov/HYSPLIT.php )

PM10 (µg/m³)

DWD AABU (Aachen) 205 m EIFE (Simmerath) 572 m date Weather type PP (hPa) TT (°C) UU (%) RR (mm) FF (m/s) DD (°) PP (hPa) TT (°C) UU (%) RR (mm) FF (m/s) DD (°) 29.01.2011 1 994 -2,8 85,1 0,1 1,5 58

23

128

3.4. Part 4: Additional Features complementing the web mapping system Hendrik Merbitz RWTH Aachen University, Department of Geography [email protected] 3.4.1. Forecast tool for PM10 levels A statistical analysis was performed, in order to assess the average PM10 concentrations 2008-2010 at background stations and traffic or industrial stations depending on the weather type (Fig. 1). The objective weather type classification from the German weather service (DWD) was chosen.

Figure 1. Average distribution of PM10 station values depending on weather type for traffic/industrial stations (above) and background stations (below)

For each day the forecasted weather type for the current day, the day after and the day after the day after are extracted from the database of the DWD. For the 3 weather types graphs are produced showing the range (min to max) and the median PM10 concentration for all background sites and for all traffic or industrial sites (Fig. 2).

129

Figure 2. Example for statistical forecast of PM10 levels at background and traffic/industrial sites.

130

3.4.2. Back-Trajectories As an additional layer on the Web Map Tool 48-hours back trajectories are calculated with the HYSPLIT 4 model (Draxler & Hess 1997) in order to locate the origin and transport path of advected air masses. Fig. 3 shows an example for back trajectories calculated for the EMR-tripoint B/NL/D.

Figure 3. Back-trajectories.

3.4.3. References Draxler, R.R. & Hess, G.D. (1997): Description of the HYSPLIT_4 Modeling System. NOAA Technical Memorandum ERL ARL-224. Air Resources Laboratory Silver Spring, Maryland, December 1997 (revised 2010, December). http://ready.arl.noaa.gov/HYSPLIT.php Accessed: 01.10.2012

131

3.5. Part 5: Additional Features complementing the web mapping system Modeled Inversion Intensity Gunnar Ketzler RWTH Aachen University, Department of Geography [email protected] 3.5.1. Modeled Inversion Intensity – general information During daytime, the atmospheric conditions for the dispersal of emissions normally are unstable or neutral. With respect to PM, this means that a given amount of particles is diluted in greater air volumes and the PM concentration decreases in time. Under stable atmospheric conditions, which take place in a variable percentage per year, dispersion is reduced and low emission amounts can lead to high concentrations. Typically, stable conditions are coupled with ground inversions which often take place during night time. Their frequency cannot be given in general terms, but for the more prominent valley situation in Aachen this is the case in 25% - 40% of the nights (Ketzler, 2002). The time fraction of these situations during the whole year is relatively small, but as the pollutant concentrations in this time can be very high it of great importance.

Several model runs with the cold air model KLAM_21 of the German Weather Service (Sievers, 2005) were performed to evaluate the inversion characteristics in the Euregio. The model runs were performed with the standard model configuration and a standard urban warming value as described in Sievers (2005). One of the two spatial input variables is land use given by the CORINE data set as it is used for the other project tasks (see “Part 1: Development of an interpolation model for the PM10 concentrations in the EMR”). The CORINE land use classes were transferred into the KLAM_21 input classes using a transfer table (see annex). The other input variable is a digital elevation model (DEM) generated from SRTM radar data (USGS, 2010).

132

3.5.2. Typical evening cooling Local inversions mostly occur as a result of cooling near the ground especially during evening and night. Evening and night time cooling is a typical phenomenon on days with clear sky conditions. On such days, the earth´s surface loses energy by long-wave radiation (IR) and the air near the ground will slowly cool down while the air above is not affected. The intensity of cooling near the ground (here at 2 m AGL) is one main information to evaluate the intensity of ground inversions and, thus, the risk of accumulation of air pollutants. The main KLAM_21 output variable for this analysis is energy deficit data for each raster point. From these data, an evening cooling value for the time +3h from after sunset was calculated using a cooling energy constant of 830 Jm-3K-1 (Ketzler, 1997). This evening cooling model result for +3 from sunset under standard conditions is the “typical evening cooling [°C]” used for the web mapping tool. The results show differences between hill tops and urban areas as relatively warm and, thus, with a lower inversion intensity. Cold air will flow down the hills and will be replaced by warmer air from above, in cities cold air will be warmed and lifted up due to urban warming. In non-urbanized valley bottoms, cold air accumulates, which leads to very strong evening cooling and high inversion intensity respectively. In mediums sized valleys especially in the direct surroundings of the Eifel / Ardennes low mountain area, evening cooling reaches highest intensities. In areas with intense evening cooling emissions should be reduced with high priority as they can lead to severe pollutant concentrations. However, these areas can have ventilation functions for downstream regions with higher pollutant concentrations and the conditions for air flow should be left as good as possible as well. 3.5.3. References Ketzler, G. (1997): Vergleichende Untersuchungen der Temperaturverhältnisse in Städten unter besonderer Berücksichtigung der Temperaturänderungsraten Ketzler, G. (2002): Untersuchungen an Kaltluftströmen in kleinen stadtnahmen Tälern; in: Aachener Geographische Arbeiten, 36 SIEVERS, U. (2005): Das Kaltluftabflussmodell KLAM_21. Theoretische Grundlagen, Anwendung und Handhabung des PC-Modells. Berichte des Deutschen Wetterdienstes 227 USGS - U.S. Geological Survey (2010): SRTM Radar Data, http://nationalmap.gov/viewer.html

133

Annex Transfer table from CORINE to KLAM_21 land use classes CLC_CODE

LABEL1

LABEL2

LABEL3

KLAM_21

111

Artificial surfaces

Urban fabric

Continuous urban fabric

1

112

Artificial surfaces

Urban fabric

Discontinuous urban fabric

2

121

Artificial surfaces

Industrial or commercial units

5

122

Artificial surfaces

123

Artificial surfaces

124

Artificial surfaces

131

Artificial surfaces

132

Artificial surfaces

133

Artificial surfaces

141

Artificial surfaces

142

Artificial surfaces

211

Agricultural areas

Industrial, commercial and transport units Industrial, commercial and transport units Industrial, commercial and transport units Industrial, commercial and transport units Mine, dump and construction sites Mine, dump and construction sites Mine, dump and construction sites Artificial, nonagricultural vegetated areas Artificial, nonagricultural vegetated areas Arable land

212

Agricultural areas

213

Road and rail associated land

networks

and

8

Port areas

8

Airports

8

Mineral extraction sites

7

Dump sites

7

Construction sites

4

Green urban areas

6

Sport and leisure facilities

4

Non-irrigated arable land

7

Arable land

Permanently irrigated land

7

Agricultural areas

Arable land

Rice fields

7

221

Agricultural areas

Vineyards

6

222

Agricultural areas

Fruit trees and berry plantations

6

223

Agricultural areas

Olive groves

6

231

Agricultural areas

Permanent crops Permanent crops Permanent crops Pastures

Pastures

7

241

Agricultural areas

Annual crops associated permanent crops

242

Agricultural areas

243

Agricultural areas

Heterogeneous agricultural areas Heterogeneous agricultural areas Heterogeneous agricultural

with

7

Complex cultivation patterns

6

Land principally occupied by agriculture, with significant areas of

7

134

areas

natural vegetation Agro-forestry areas

3

semi

Heterogeneous agricultural areas Forests

Broad-leaved forest

3

semi

Forests

Coniferous forest

3

semi

Forests

Mixed forest

3

semi

Natural grasslands

7

Moors and heathland

6

Sclerophyllous vegetation

6

Transitional woodland-shrub

6

Beaches, dunes, sands

7

Bare rocks

7

Sparsely vegetated areas

7

Burnt areas

7

Glaciers and perpetual snow

9

Inland marshes

7

Peat bogs

7

Salt marshes

7

Salines

7

Intertidal flats

7

244

Agricultural areas

311

Forest and natural areas Forest and natural areas Forest and natural areas Forest and natural areas

312 313 321

322

Forest and natural areas

semi

323

Forest and natural areas

semi

324

Forest and natural areas

semi

331

Forest and natural areas

semi

332

Forest and natural areas

semi

333

Forest and natural areas

semi

334

Forest and natural areas

semi

335

Forest and natural areas

semi

411

Wetlands

412

Wetlands

421

Wetlands

422

Wetlands

423

Wetlands

511

Water bodies

Scrub and/or herbaceous vegetation associations Scrub and/or herbaceous vegetation associations Scrub and/or herbaceous vegetation associations Scrub and/or herbaceous vegetation associations Open spaces with little or no vegetation Open spaces with little or no vegetation Open spaces with little or no vegetation Open spaces with little or no vegetation Open spaces with little or no vegetation Inland wetlands Inland wetlands Maritime wetlands Maritime wetlands Maritime wetlands Inland waters

Water courses

9

512

Water bodies

Inland waters

Water bodies

9

521

Water bodies

Marine waters

Coastal lagoons

9

522

Water bodies

Marine waters

Estuaries

9

523

Water bodies

Marine waters

Sea and ocean

9

135

3.6. Part 6: Mobile PM Measurements in Aachen, Maastricht and Liège Gunnar Ketzler, Hendrik Merbitz RWTH Aachen University, Department of Geography [email protected], [email protected] In connection with other measurement activities in the PMLab project, mobile measurements were performed with a mixed team of PMLab-staff from Belgium, the Netherlands and Germany in the cities of Aachen, Maastricht and Liège. Additionally, students of Geography at the RWTH Aachen could take part in these activities and by doing so were able to both learn about the border crossing air quality topics and the cooperation possibilities. 3.6.1. Mobile Measurements in Aachen In a previous project (City2020+), mobile PM measurement campaigns were performed in the city of Aachen. These campaigns served as the basis of the measurements in Liège, Angleur and Maastricht. The results of the campaigns in Aachen have been published in Merbitz et al. 2012a and Merbitz et al. 2012b and show a large spatial variability of urban particulate matter (PM10, PM2.5) concentrations, mainly depending on traffic intensity and building density.

3.6.2. Mobile Measurements in Angleur Mobile Measurements of PM concentrations were performed in the town of Angleur in the vicinity of several large roads and in residential areas. Fig. 1 shows the topography in Angleur. The area of investigation is situated in a valley with differences in altitude between 200 meters affecting local climatic conditions and air quality.

Figure 1. Angleur topography (Digital elevation model: ASTER, ASTER GDEM Validation Team 2011).

136

Fig. 2 shows the land use in Angleur together with the mobile measurement sites (1 to 7) and the ISSeP station (‘ISSeP-Stat’) which served as reference site.

Figure 2. Land use in the area of Angleur.

Table 1 shows the average measured PM10 concentrations at the monitoring sites. Highest PM10 levels were observed at site number 7 in a narrow street while lowest concentrations were measured at sites 3, 4 and 6 which are situated at junctions and open squares. Traffic can be identified as major source for PM in the Angleur area during the measurements. Table 1. PM10 averages at the 7 mobile measurement sites.

Site PM10 (µg/m³) 1 28.1 2 29.9 3 25.9 4 25.9 5 27.5 6 24.7 7 30.4

137

3.6.3. Mobile Measurements in Maastricht Mobile PM measurements were carried out in the vicinity of the permanent measurement station of Province Limburg (A2 Nassaulaan) in order to estimate the impact of a large road on PM10 and PM2.5 levels in nearby residential zones. Fig. 3 shows the mobile measurement sites near the A2 and the permanent reference site operated by the Province of Limburg (‘Station’).

Figure 3. Mobile measurement sites near the A2

The A2 with an average daily traffic intensity of >47,000 vehicles per day can be clearly identified as the dominating source of PM in the area of investigation.

Figs. 4 and 5 show the concentration profile along 2 transects (sites 8-1-2-3 and sites 6-5-4). 40

8-1-2-3 6-5-4

35

PM10 (µg/m³)

7 30

25

20

15 0

50

100

150

200

Distance to A2 (m)

Figure 4. PM10 concentration profiles.

138

250

24

8-1-2-3 6-5-4

22

7 PM2.5 (µg/m³)

20

18

16

14

12 0

50

100

150

200

250

Distance to A2 (m)

Figure 5. PM2.5 concentration profiles.

Concentration profiles show a sharp decrease of PM10 and PM2.5 levels between the site closest to the A2 (site 8) and the sites with larger distance to the A2. Further decay of concentrations with distance is limited, showing rather local influences of the A2 on air quality. 3.6.4. Mobile Measurements in Liège / Centre In the inner city of Liège the impact of a large Boulevard on ambient air quality was captured by mobile measurements and dispersion modeling. The results of these measurements and model results in the city centre of Liège are published in Merbitz et al. 2012c (paper attached). 3.6.5. References ASTER GDEM Validation Team (2011): ASTER Global Digital Elevation Model Version 2 – Summary of Validation Results August 31, 2011. Merbitz, H., Buttstädt, M., Michael, S., Dott, W., Schneider, C. (2012a): GIS-based identification of spatial variables enhancing heat and poor air quality in urban areas. Applied Geography 33, 94–106. Merbitz, H, Fritz, S., Schneider, C. (2012b): Mobile measurements and regression modeling of the spatial particulate matter variability in an urban area. Science of the Total Environment 438, 389–403. Merbitz, H., Detalle, F., Ketzler, G., Schneider, C., Lenartz, F. (2012c): Small scale particulate matter measurements and dispersion modelling in the inner city of Liège, Belgium. International Journal of Environment and Pollution 50 (1/2/3/4), 234–249.

139

3.7. Part 7: Data transformation Fabian Lenartz1, Achim Knörchen2, Hendrik Merbitz2 and Gunnar Ketzler2 1 ISSeP, Department of air quality 2 RWTH Aachen University, Department of Geography [email protected], [email protected], [email protected], [email protected] 3.7.1. Data base The near-real time application, which provides PM10 concentration maps, relies on a data base running on PostgreSQL (http://www.postgresql.org/). It consists of 13 tables gathering details about the locations where measurements take place, the parameters of interest, the methods in use, the actual data, the calibration equations of the monitors with respect to the reference gravimetric method of the different networks and that of PM-Lab, the German Weather Service forecast and the time at which daylight saving occurs. The structure of the 9 tables useful for this web GIS application is depicted in Fig. 1. See appendix A for some more details about the tables.

Figure 1. Main tables of the data base

Among them, 2 are pmxmonitor_mapserver. 

of

major

importance

for

our

service:

data_mapserver

and

The first one collects PM10 records from all sites, either used by the model, or selected for illustrative purposes. It is fed by combining data_eea and data_networks, which both have the same structure, i.e. 4 fields constituting the primary key and guaranteeing uniqueness of each entry (in red for each table in Fig. 1) and 2 fields containing the actual 1- and 24-h averaged value, but which differ by their source, i.e. one originates from the EEA through IRCEL-CELINE, the other one from the ISSeP, LANUV-NRW, Province Limburg, RIVM, VMM

140

and ZIMEN networks. Both threads are managed in a similar way. A bash script, scheduled every hour, saves the latest data available, then runs an Octave script that transforms the input files so that they can be easily inserted into the data base. The formatting process simply consists in replacing station names and SAROAD codes by our own codes and in adding another one that describes the monitor in use at the considered location and time. If the measured value is provided as a 24-h averaged value it will appear in the last field of the prepared file and the second to last field is left with an exclusion value (-9999); if the measured values is provided as a 1-h averaged value it will appear in the second to last field of the prepared file and the 24-h averaged value will be computed on the basis of the 23 previous records. Hence a typical input file, e.g exported from the LANUV-NRW website/teletext, which looks like this DENW058;81102;2013030711;45 DENW059;81102;2013030711;30 DENW062;81102;2013030711;43 DENW064;81102;2013030711;18 DENW066;81102;2013030711;43 DENW074;81102;2013030711;55 DENW094;81102;2013030711;23 DENW096;81102;2013030711;32 DENW100;81102;2013030711;34 DENW182;81102;2013030711;37 DENW207;81102;2013030711;44 DENW259;81102;2013030711;48 DENW3XX;81102;2013030711;48

will be processed into this 1;20;7;2013-03-07 10:00:00;-9999.0;23.0 1;21;7;2013-03-07 10:00:00;-9999.0;44.0 1;23;7;2013-03-07 10:00:00;-9999.0;55.0 1;25;7;2013-03-07 10:00:00;-9999.0;18.0 1;26;13;2013-03-07 10:00:00;-9999.0;43.0 1;30;13;2013-03-07 10:00:00;-9999.0;45.0 1;32;3;2013-03-07 10:00:00;-9999.0;30.0 1;36;11;2013-03-07 10:00:00;-9999.0;34.0 1;37;7;2013-03-07 10:00:00;-9999.0;48.0 1;38;7;2013-03-07 10:00:00;-9999.0;32.0 1;39;7;2013-03-07 10:00:00;-9999.0;43.0 1;56;7;2013-03-07 10:00:00;-9999.0;37.0 1;57;7;2013-03-07 10:00:00;-9999.0;48.0.

This multiple-source approach aims at minimizing the risk of empty or missing records, due to a potential problem that may affect one or several of our data providers, hence allows us to generate and serve high resolution maps of PM10 concentration in a more (though not perfectly) reliable way. See appendix B for some more details about the scripts. 

The second one stores the calibration coefficients determined through equivalence campaigns led, on one hand by the networks, and on the other hand by the PM-Lab Monitoring team of our consortium. With this information we can transform, in a first step, the measurement back into the raw values from the analyzer, and in second step, forth into

141

the PM-Lab calibrated values. This is performed in /home/postgres/grassdata/pm10_model.sql and pm10_stations.sql files.

both

the

3.7.2. SQL queries In order to feed the web application, two SQL queries are performed every hour to retrieve data and transform them. This computation is performed in two steps: - First, data from the different networks are transformed back into raw values, as originally measured by the analyzers; - Secondly, they are transformed into the so-called PM-Lab values.

Although pretty close to the original calibration factors (slope and offset), the ones determined during PM-Lab project provide somewhat more homogeneous maps over the EMR. Both queries are provided in Appendix C.

142

Appendix A – Data base Here are presented two of the data base tables: data_networks and pmxmonitor_mapserver, which respectively contains the different network measurements that arrive continuously from their website and the calibration coefficients used by the networks as well as those determined by the PM-Lab consortium. data_networks c_parameter

c_location

c_method

time

value

valrunavg

1

1

9

10/07/2013 01:00

-9999

21.8

1

2

9

10/07/2013 01:00

-9999

17.8

1

3

9

10/07/2013 01:00

-9999

18.3

1

4

9

10/07/2013 01:00

-9999

-9999

1

5

9

10/07/2013 01:00

-9999

24.8

1

6

9

10/07/2013 01:00

-9999

25.8

1

8

9

10/07/2013 01:00

-9999

16.3

1

9

9

10/07/2013 01:00

-9999

24.3

1

10

9

10/07/2013 01:00

-9999

22.3

1

12

9

10/07/2013 01:00

-9999

23.3

1

13

9

10/07/2013 01:00

-9999

-9999

1

14

4

10/07/2013 01:00

-9999

26.2

1

15

4

10/07/2013 01:00

-9999

26.9

1

16

4

10/07/2013 01:00

-9999

25

1

17

4

10/07/2013 01:00

-9999

29.4

1

18

11

10/07/2013 01:00

-9999

35

1

20

7

10/07/2013 01:00

-9999

21

1

21

7

10/07/2013 01:00

-9999

27

1

23

7

10/07/2013 01:00

-9999

24

1

25

7

10/07/2013 01:00

-9999

17

1

26

13

10/07/2013 01:00

-9999

21

1

30

13

10/07/2013 01:00

-9999

31

1

32

3

10/07/2013 01:00

-9999

20

1

36

11

10/07/2013 01:00

-9999

18

1

37

7

10/07/2013 01:00

-9999

26

1

38

7

10/07/2013 01:00

-9999

23

1

39

7

10/07/2013 01:00

-9999

21

143

1

40

7

10/07/2013 01:00

-9999

17

1

45

3

10/07/2013 01:00

-9999

-9999

1

46

3

10/07/2013 01:00

-9999

-9999

1

47

3

10/07/2013 01:00

-9999

-9999

1

48

3

10/07/2013 01:00

-9999

-9999

1

51

9

10/07/2013 01:00

-9999

-9999

1

54

3

10/07/2013 01:00

-9999

-9999

1

56

7

10/07/2013 01:00

-9999

21

1

57

7

10/07/2013 01:00

-9999

32

1

1

9

10/07/2013 02:00

-9999

22.8

1

2

9

10/07/2013 02:00

-9999

20.3

1

3

9

10/07/2013 02:00

-9999

17.8

1

4

9

10/07/2013 02:00

-9999

-9999

1

5

9

10/07/2013 02:00

-9999

25.8

1

6

9

10/07/2013 02:00

-9999

23.3

1

8

9

10/07/2013 02:00

-9999

14.3

1

9

9

10/07/2013 02:00

-9999

38.8

1

10

9

10/07/2013 02:00

-9999

24.8

1

12

9

10/07/2013 02:00

-9999

23.8

1

13

9

10/07/2013 02:00

-9999

-9999

1

14

4

10/07/2013 02:00

-9999

20

1

15

4

10/07/2013 02:00

-9999

27.5

1

16

4

10/07/2013 02:00

-9999

33.1

1

17

4

10/07/2013 02:00

-9999

31.9

1

18

11

10/07/2013 02:00

-9999

49

1

20

7

10/07/2013 02:00

-9999

26

1

21

7

10/07/2013 02:00

-9999

44

1

23

7

10/07/2013 02:00

-9999

28

1

25

7

10/07/2013 02:00

-9999

23

1

26

13

10/07/2013 02:00

-9999

30

1

30

13

10/07/2013 02:00

-9999

33

1

32

3

10/07/2013 02:00

-9999

28

1

36

11

10/07/2013 02:00

-9999

25

1

37

7

10/07/2013 02:00

-9999

27

144

1

38

7

10/07/2013 02:00

-9999

24

1

39

7

10/07/2013 02:00

-9999

21

1

40

7

10/07/2013 02:00

-9999

17

1

45

3

10/07/2013 02:00

-9999

23

1

46

3

10/07/2013 02:00

-9999

25

1

47

3

10/07/2013 02:00

-9999

24

1

48

3

10/07/2013 02:00

-9999

21

1

51

9

10/07/2013 02:00

-9999

-9999

1

54

3

10/07/2013 02:00

-9999

17

1

56

7

10/07/2013 02:00

-9999

-9999

1

57

7

10/07/2013 02:00

-9999

27

1

1

9

10/07/2013 03:00

-9999

24.3

1

2

9

10/07/2013 03:00

-9999

25.3

1

3

9

10/07/2013 03:00

-9999

18.3

145

pmxmonitor_mapserver c_param eter

c_loca tion

c_met hod

time_monitor _start

time_monit or_end

network_o ffset

network_ slope

pmlab_of fset

pmlab_s lope

1

1

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

2

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

3

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

4

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

5

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

6

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

8

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

9

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

10

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

11

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

12

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

13

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

-2.73

1

1

14

4

01/01/2012 00:00

01/01/2015 00:00

0

1.25

0

1.266

1

15

4

01/01/2012 00:00

01/01/2015 00:00

0

1.25

0

1.266

1

16

4

01/01/2012 00:00

01/01/2015 00:00

0

1.25

0

1.266

1

17

4

01/01/2012 00:00

01/01/2015 00:00

0

1.25

0

1.266

1

18

11

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

20

7

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

21

7

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

23

7

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

25

7

28/03/2012 00:00

01/01/2015 00:00

0

1

0

1

146

1

26

13

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

30

13

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

32

3

01/01/2012 00:00

01/01/2015 00:00

0

1.29

0

1.266

1

36

37

19/04/2012 00:00

01/01/2015 00:00

0

1

0

1

1

37

7

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

38

7

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

39

7

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

40

7

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

42

6

01/01/2012 00:00

01/01/2015 00:00

0

1

-3.36

0.903

1

43

6

01/01/2012 00:00

01/01/2015 00:00

0

1

-3.36

0.903

1

45

3

01/01/2012 00:00

01/01/2015 00:00

2.2

1.2

0

1.266

1

46

3

01/01/2012 00:00

01/01/2015 00:00

2.2

1.2

0

1.266

1

47

3

01/01/2012 00:00

01/01/2015 00:00

2.2

1.2

0

1.266

1

48

3

01/01/2012 00:00

01/01/2015 00:00

2.2

1.2

0

1.266

1

51

9

01/01/2012 00:00

01/01/2015 00:00

-2.729

1

2

0.695

1

54

3

01/01/2012 00:00

01/01/2015 00:00

2.2

1.2

0

1.266

1

55

6

01/01/2012 00:00

01/01/2015 00:00

0

1

-3.36

0.903

1

56

37

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

1

57

7

01/01/2012 00:00

01/01/2015 00:00

0

1

0

1

147

Appendix B – Scripts Here are some of the scripts used to format and insert data into the data base. The main one insertdata.sh is run every hour by the scheduler at HH:10. It calls, for each source i.e. EEA, IRCELCELINE, LANUV-NRW, RIVM, PLIM and ZIMEN, an Octave script f[networkname]2db.m that actually reads the file that is retrieved from the network website and format the entry for the database. Afterwards, it calls a shell script insert[networkname]data.sh that inserts what has just been prepared into the database. All scripts are located in the /home/pmlu/data/scripts/ directory. insertdata.sh #!/bin/bash rep=/home/pmlu/data/scripts/ # EEA date >> ${rep}logeea octave < ${rep}feea2db.m >> ${rep}logeea 2>&1 sh ${rep}inserteeadata.sh 2>> ${rep}logeea # IRCEL-CELINE date >> ${rep}logirceline octave < ${rep}firceline2db.m >> ${rep}logirceline 2>&1 sh ${rep}insertircelinedata.sh 2>> ${rep}logirceline # LANUV-NRW date >> ${rep}loglanuv octave < ${rep}flanuv2db.m >> ${rep}loglanuv 2>&1 sh ${rep}insertlanuvdata.sh 2>> ${rep}loglanuv # RIVM date >> ${rep}logrivm octave < ${rep}frivm2db.m >> ${rep}logrivm 2>&1 sh ${rep}insertrivmdata.sh 2>> ${rep}logrivm # PLIM date >> ${rep}logplim octave < ${rep}fplim2db.m >> ${rep}logplim 2>&1 sh ${rep}insertplimdata.sh 2>> ${rep}logplim # ZIMEN date >> ${rep}logzimen octave < ${rep}fzimen2db.m >> ${rep}logzimen 2>&1 sh ${rep}insertzimendata.sh 2>> ${rep}logzimen

flanuv2db.m clear all close all clc tic % FORMATTING LANUV DATA FOR OUR DATABASE [v1,v2,v3,v4]=textread('/home/pmlu/data/lanuv/lanuv.dat','%s%s%s%s','delimiter',';' ); STATION=v1; POLCODE=sscanf(sprintf('%s*',v2{:}),'%f*'); for HH=1:length(v3) HOUR(HH,:)=char(v3(HH)); end VALUE=sscanf(sprintf('%s*',v4{:}),'%f*'); clear v1 v2 v3 v4 HH % Transform time into a number (Matlab format) DATENUMBER=NaN*ones(size(HOUR,1),1); JUSTHOUR=NaN*ones(size(HOUR,1),1); for kt=1:size(HOUR,1)

148

DATENUMBER(kt,1)=datenum(str2double(HOUR(kt,1:4)), str2double(HOUR(kt,5:6)), str2double(HOUR(kt,7:8)), str2double(HOUR(kt,9:10)), 0, 0); JUSTHOUR(kt,1)=str2double(HOUR(kt,9:10)); end % Sort data (in the following order for the 33 stations) los={'DENW094','DENW207','DENW074','DENW064','DENW062', ... 'DENW058','DENW059','DENW100','DENW259','DENW096', ... 'DENW066','DENW182','DENW329'}; cost=[20;21;23;25;26; ... 30;32;36;37;38; ... 39;56;57]; com=[7;7;7;7;13; ... 13;3;11;7;7; ... 7;7;7]; % Loop over the stations fid=fopen('/home/pmlu/data/lanuv/lanuv_r4db.dat','w'); for ks=1:length(los) disp(['Station considered: ' char(los(ks))]); ind=find(strcmp(STATION,char(los(ks)))==1); if isempty(ind) % This part should be useless disp(['We do not have data for ' char(los(ks))]); datenumber=DATENUMBER(1); value=-9999; else % This part is the important one disp(['We do motherfucking have data for ' char(los(ks))]); datenumber=DATENUMBER(ind); value=VALUE(ind); end if (datenumber>datenum(2012,3,25,2,0,0))&&(datenumberdatenum(2013,3,31,2,0,0))&&(datenumberdatenum(2014,3,30,2,0,0))&&(datenumberdatenum(2015,3,29,2,0,0))&&(datenumber ${rep}all_r4db.dat psql -d pmxdb -f ${rep}scripts/insertionNET done < ${rep}lanuv/lanuv_r4db.dat

As the scripts for the other networks are very similar, we chose only to display those related to LANUV-NRW.

149

Appendix C – SQL queries Here is the SQL query retrieving data from all stations displayed in the web application: /home/postgres/grassdata/pm10_stations.sql. pm10_stations.sql SELECT data_networks.time, data_networks.c_location, location.short_name_irceline, location.short_name, location.long_name_html, location.network_in_charge, location.longitude, location.latitude, location.use_in_the_model, CASE WHEN data_networks.valrunavg IS NULL THEN -9999 ELSE CASE WHEN data_networks.valrunavg=-9999 THEN -9999 ELSE ((data_networks.valrunavgpmxmonitor_mapserver.network_offset)/pmxmonitor_mapserver.network_slope)*pmxmonitor _mapserver.pmlab_slope+pmxmonitor_mapserver.pmlab_offset END END as valrunavg FROM data_networks RIGHT JOIN location ON (data_networks.c_location = location.code_location) AND data_networks.time = date_trunc('hour', now()) - interval '5 hour' LEFT JOIN parameter ON (data_networks.c_parameter = parameter.code_parameter) AND data_networks.c_parameter = 1 LEFT JOIN pmxmonitor_mapserver ON (pmxmonitor_mapserver.c_parameter=data_networks.c_parameter and pmxmonitor_mapserver.c_location=data_networks.c_location and pmxmonitor_mapserver.c_method=data_networks.c_method and data_networks.time>=pmxmonitor_mapserver.time_monitor_start and data_networks.time50t/a) are directly located in the EMR (see Figure ), but there are a lot of emitting facilities close to the EMR, especially in the Rhenish Lignite Mining Area (DE). To assure that such closely related industrial sources are considered in the analysis, the “administrative” boundaries of the EMR were extended for 25km in each direction. On the other side, the extension doesn’t go too far into or close to other agglomerations (e.g. Brussels, Cologne (partially included) or Düsseldorf).

Figure 3. Location of Industrial Point Soruces Source: own draft (further sources see figure)

157

From 2001 to 2010 a very interesting development of PM10 point sources can be observed in the (extended) EMR. In general, the development of facility numbers, total emissions and the emissions per facility shows a declining “main-trend”. But within this main trend there are some fluctuations which build a “sub-trend”. As can be seen from Figure , the number of reporting facilities and the PM10 emissions from industrial point sources in the EMR declined more or less continuously from 2001 to 2010 (“main-trend”). But especially from 2007 to 2008 (+1.531t) and also from 2009 to 2010 (+700t) significant increases of PM10 emissions can be observed (“sub-trend”). In 2008, the reference year for the PM-Lab project, 5 facilities which are located directly in the EMR reported to the E-PRTR. Together they had PM10 emissions of 2.855t. Looking at the extended area (EMR+25km, see Figure ), the observed trends look quite similar, but in most cases the emission changes are less dynamic than in the EMR. This can be best explained with a look on the relevant sectors and the geographical distribution of the point sources (see Figure ).

Figure 4. General development of Industrial Point Sources Source: E-PRTR Database; own calculations and draft

The spatial and sectoral distribution of point sources over time shows also some interesting main trends and patterns. Starting in 2001, the main sources were mostly located in the Province Liège and came from the metal producing and processing sector (mainly plants of ArcelorMittal, see Case studies) and the mineral industry. Other important sources are the coal fired power plants and coal processing plants of the German energy supplier RWE, which mostly lie outside the boundaries of the EMR, except the Power Plant Weisweiler. This “bi-polar” general pattern is constant over the time. Compared to the before mentioned, some other point sources have rather low PM10 emissions or disappear completely over time (e.g. 3B-Fibreglass, Chemelot, Dumont Wautier, CBR, O-I, Electrabel Awirs), while others appear in single years or reoccur from time to time (e.g. E.ON, Knauf Insulation). The next section will present the results of the case studies (in alphabetical order) and gives an insight into the reasons, which lie behind the presented developments.

158

Figure 5. Development of PM10 Emissions from Industrial Point Sources in Space and Time Source: own draft (further sources see figure)

159

3.8.3.2. Case Studies - Industrial Point Sources The influences of companies PM10 emissions and reduction activities are very diverse and therefore existing governance approaches, too. These might be categorized as following:

1. Informal governance (mostly voluntary)  company strategies (general strategy/ decision making, e.g. raising and lowering of capacities, reactions to the (market) environment or business cycles)  sustainability strategies (defining the companies attitude towards environmental actions)  environmental management systems 2. Formal governance (mostly by legislation)  European environmental legislation and corresponding national legislation, e.g. IPPC directive (since 1996) and European Ambient Air Quality Directive (2008/50/EC) (motivation for reduction measures, e.g. dust filters) As the interviews showed, companies as well as authorities mentioned higher pressure regarding PM10 reduction measures, especially through the revision of the European Ambient Air Quality Directive (2008/50/EC). A relation of this directive to industry is made more or less implicitly, but very perceptible for the actors. The directive demands a coordination of air quality plans and the application of other regulations23. Especially a consideration of the air quality targets during the process of environmental permitting in the sense of the IPPC directive is obliged. So, stricter limit values for ambient air can lead to stricter environmental requirements for individual companies. For a better interpretation of the case studies below, a brief summary of the IPPC-process is necessary24. In principle, the IPPC directive obliges industrial and also agricultural facilities with high pollution potential to apply for environmental permits. As precondition for a permit, companies have to consider certain basic obligations (e.g. use of “best available techniques” (BAT), waste management, energy efficiency measures, site restoration, and accident/ incident management). In addition there are additional (individual) requirements that for example include limit values for certain pollutants and corresponding monitoring, prevention and reduction measures. Dust (here total dust!) is mentioned as one important pollutant in annex III.

23

Directive 2008/1/EC of the European Parliament and of the Council of 15 January 2008 concerning integrated pollution prevention and control (IPPC); Note: this directive was revised and recently replaced by the Directive on industrial emissions 2010/75/EU (IED) 24 see for example: http://europa.eu/legislation_summaries/environment/air_pollution/l28045_en.htm

160

3-B Fibreglass 3-B Fibreglass is a rather new company, which was founded in 2008, after its former parent company Owens Corning (USA) had to sell the plant at Battice (together with another in Norway) due to regulatory concerns of the European Commission that were associated with Owens Corning’s acquisition of parts of Saint-Gobain Vetrotex. Since 2012 the company is owned by the Indian Braj Binani Group25. Today, approximately 450 people work in the 3-B plant at Battice that mainly produces glass fibres26. PM10 emissions were only reported to the E-PRRT/ EPER in 2001 (89,5t) under the company’s former name Owens Corning Composites SPRL. This indicates that the foundations for the environmental improvements were already built under the ownership of Owens Corning in the early 21st century. One main reason for the companies’ dust emissions were the use of boron and fluorides in the glass production. Therefore a combination of new combustion technologies (oxygen-gas firing and furnace design) and a new, boron and fluoride free glass formulation lead (among other improvements) to a dust level reduction of approx. 90% and specific dust emissions on BAT level27. ArcelorMittal Liège ArcelorMittal has been the company with the highest PM10 emission in the EMR over years. The high PM10 emissions are a result of the company’s “Upstream” section activities at Liège with its production sites at Ougrée, Seraing (“Coke-Fonte”) and Chertal. In fact, the Coke/ Fonte part of the upstream section of ArcelorMittal Liège has been the biggest pollutant in the EMR for long time, followed by the plant at Chertal, which had the second highest PM10 emissions in the beginning of the 21st century. Explanations for the development of the company’s PM10 emissions can be derived from a closer look at the company’s younger history: In 2003 the company decided to concentrate its future investments for the renovation of blast furnaces in Europe mainly to its most profitable coastal locations28, what implicates that no major investments in locations like Liège were planned. Independently of this, the production at Liège continued in the following years, but was often and rapidly adapted to the development of global steel demand, especially by the (de-) commission of the blast furnaces at Seraing and/or Ougree. In 2005 one of the blast furnaces was decommissioned and should be restarted in 200729, what may explain the relatively low PM10 emissions in 2007 compared to the years before. After negotiations with the federal and regional government about CO2 emission allowances, the blast furnace at Seraing (HF6) started to produce again in beginning of 200830. The decline from 2008 to 2009 can also be explained by the temporally decommission of both blast furnaces in mid of 2008. One of them was re-launched in 2010, what could explain the PM10 value for that year31. Finally, ArcelorMittal decided to permanently close down the upstream section (blast furnaces at Ougrée and Seraing; foundry at Chertal) by the end of 2011, while other plants will continue their production 32.

25

Owens Corning (2012): Owens Corning Completes Sale of Composite Manufacturing Plants in Belgium and Norway. http://owenscorning.mediaroom.com, 01.05.2008/ 06.12.2012; 3b fibreglass (2012a): www.3b-fibreglass.com/ourcompany/history/, 06.12.2012 26 economie.wallonie (2011): www.economie.wallonie.be, 01.12.2011 27 3b fibreglass (2012b): Advantex – Green Solution for the Blue Planet. www.3b-fibreglass.com, 06.12.2012; kunststoffFORUM (2001): Umweltanstrengungen von Owens Corning im Rampenlicht von EU-Konferenz. www.kunststoffforum.de; 03.12.2003/ 28.11.2011; Owens Corning (2004): Owens Corning – Usine de Battice Implémentation des “BAT’s” (presentation from 30.09.2004) 28 ArcelorMittal (2011a): www.arcelormittal.com, About > History > History of Arcelor, 15.11.2011 29 ArcelorMittal (2011b): www.arcelormittal.com, About > History > History of Arcelor > Arcelor Highlights, 15.11.2011 30 ArcelorMittal (2011c): www.arcelormittal.com, > About > History > 2008 highlights, 15.11.2011 31 L'essentiel Online (2011): Endgültiges Aus für Hochöfen in Lüttich. www.lessentiel.lu, 16.11.2011 32 flanderninfo.be (2011): “ArcelorMittal schockt seine Mitarbeiter”, www.deredactie.be, 13.10.2011

161

CBR SA - site de LIXHE33 CBR is a cement manufacturer, which is part of the German enterprise HeidelbergCement AG. The production site at Lixe lies in the north of the Province de Liège and employs appr. 350 people 34. Like Dumont Wautier SA (see below), CBR also reported only in 2001 PM10 emission to the EPER / E-PRTR and also with a high amount of 480t. In the following years, CBR did a lot of improvements, which led to much lower PM10 values which now lay around 36t. Major improvements were made during a complete change of the production process in 2001 that was part of a modernisation and restructuring scheme35: In a first step, the capacity of the existing dry kiln was raised and later the wet kiln at Lixhe was closed. In addition, the remaining kiln was equipped with bag filters that replaced the existing electrostatic precipitators, so that the new kiln has a more effective dust reduction. CBR takes also part in the “Plan Pics de Pollution” (see later), which means that it has to adapt certain production processes if pollution peaks occur.

Chemelot Site Permit The Chemelot Site Permit B.V. is a service institution which takes over the coordination of safety, health and environment activities of all companies that are located at the chemical production complex Chemelot at Geleen (NL). This also includes demands for permits as well as reporting to the E-PRTR36. PM10 was only reported to the EPER/ E-PRTR in 2001 and 2004. The first was done by the company DSM itself, while later on Chemelot/ Chemelot Site Permit are mentioned in the E-PRTR database. Therefore it can be assumed that still DSM or respectively the new owner SABIC is the main PM10 source at the Chemelot complex. As can be seen from the Netherlands PRTR37, besides some fluctuations the PM10 situation generally improved over time and was always below the EPRTR benchmark after 2005 (no E-PRTR reporting). Referring to a short telephone call with Chemelot Site Permit by the end of 2011, the reductions are mainly based on technological improvements. Dumont Wautier SA (carrieres et fours a chaux) The company Dumont Wautier SA employs appr. 300 people38. It is located in the Province Liège and is part of the LHOIST group. As the name of the company indicates, it belongs to the mineral industry and focuses on quarrying of limestone and manufacturing of lime. PM10 emissions were only reported to the EPER/ E-PRTR in 2001, but with a relatively high amount of 293t. From a later environmental impact study39 it can be derived that several measures may be responsible for the positive development of the PM10 emissions: use of bag filters in the lime kilns, coverage and washing of trucks, transportation on inland waterways and railways, fuel substitution, road pavement and cleaning and a reduction plan for diffuse emissions. Electrabel Electrabel is a Belgian energy supplier, which formerly had two plants that were important PM10 sources in the EMR: Awirs and Langerloo. While the first isn’t an important PM10 source anymore, the latter - that now belongs to E.ON Benelux (see E.On Langerloo) - still is. The plant at AWIRS reported only in 2001 (163t) and 2004 (152t) to the EPER/ E-PRTR. Explanations for the PM10 development of both (!) plants can be derived from Electrabel’s general development of dust emissions and the company’s air quality ambitions, which include beside other measures the 33

besides further mentioned sources this part is based on an interview with the company economie.wallonie (2011): www.economie.wallonie.be, 01.12.2011 35 HeidelbergCement AG (2000): Annual Report 2000, p.46; HeidelbergCement AG (2002): Geschäftsbericht 2002, p. 46 36 Sabic (2005): Annual SHE Progress Report 2005. Sabic Europe B.V., p.14; Chemelot (2012): www.chemelot.nl, History; 05.12.2012 37 see RIVM (2013): www.emissieregistratie.nl/erpubliek/erpub/facility.aspx, 04.12.2013 38 economie.wallonie (2011): www.economie.wallonie.be, 01.12.2011 39 Dumont Wautier (2009): Etdue d’incidences sur l’environnement, p.26 34

162

replacement old coal-fired power stations with highly efficient natural gas-fired units and the improvement of the efficiency of the electrostatic precipitators in coal-fired power stations40. For example, one unit in the Awirs plant was converted from a conventional to a fully biomass fired one (80MW) in 2005. It can be assumed that during the conversion of the unit also improved dust reduction technologies (on BAT level) were installed that had positive effects on the “PM10 performance”41. Engineering Steel Belgium (ESB)42 Until 2009 Engineering Steel Belgium’s (ESB) electric steel plant at Seraing (Province Liège ) belonged to the US Ellwood Group and was known as Ellwood Steel Belgium. In 2009 it was purchased by the German Georgsmarienhütte Holding (GMH) GmbH and got its new name “ESB” 43. Especially the constant PM10 values from 2004 to 2008 (all 85,6t) indicate that the company produced at its capacity limits and - because of low environmental standards - very polluting, compared to the number of 118 employees 44. Not surprisingly, there were already discussions about the environmental impact of the company before the purchase by the GMH GmbH, especially about air quality and noise emissions in the night45. Because of the low global steel demand, also the GMH Group wasn’t able to invest into environmental technologies directly after the purchase, but mid 2012 the desired adaptions could take place at ESB 46. In sum approximately 25 Mio. € should be invested into environmental technologies, while the production capacity and the number of employees should be doubled in near future47. Regarding PM10 especially the implementation of a new de-dusting system is important, which will lead to specific dust emissions of 5mg/m³ 48. E.ON Langerlo49 In 2009 the plant at Langerlo was transferred from Electrabel to the German company E.ON and is since then one of the company’s two power plants in Belgium (the other one is located at Vilvoorde). The plants units use hard coal, biomass as well as natural gas that together have a capacity of 556 MW electrical power. Currently 130 people are employed at the Langerlo Powerplant. Regarding the E-PRTR PM10 emission data, the plant occurs relatively late in the sample compared to other companies: in 2008 (76,4t) and again in 2010 (55,5t). The E-PRTR value for 2008 and the recently published value for 2010 are both equal to the total dust emissions that are available from the Flemish emission register “IMJV”50 (Integraal Milieujaarverslag; value for 2010 delivered by the company/ not yet published in the IMJV). Knauf Insulation Knauf Insulation’s plant at Visé lies directly at the border of the Walloon Region with the Netherlands51. The plant is specialized in the production of glass wool and employs 300 people; the annual turnover between 2008 and 2010 was appr. 105 Mio. € 52. After reporting PM10 emissions to the E-PRTR/ EPER in 2001 (63,8t), Knauf Insulation (formerly: Knauf Alcopor) just reappeared in 2010, with even higher emissions (94,1t). Therefore, a further observation of Knauf’s latest and future activities is very interesting: In November 2011 the company decided to extend its activities at Visé, which will result in Europe’s biggest glass wool plant, with a capacity of 120.000t glass wool, compared to 90.000t today. Besides the capacity extension, also the eco-performance of the 40

Electrabel (2011): www.electrabel.be/sustainable_development, 22.11.2011 Electrabel (2005): Environemtal Report 2005, p. 14 this part is also based on an interview with the company 43 Engineering Steel Belgium (2012): www.esb.be/uk/company/history.html, 06.12.2012 44 economie.wallonie (2011): www.economie.wallonie.be, 01.12.2011 45 LeSoir.be: “Défense à ESB de polluter”, 05.11.2008 46 Engineering Steel Belgium (2012) 47 LeSoir (2011b): GMH veut doper l’aciérie de Seraing. http://archives.lesoir.be, 12.07.2011 48 LeSoir (2011a): ESB promet une usine moins polluante. http://archives.lesoir.be, 23.06.2011 49 E.ON (2013): Locatie Langerlo (Factsheet); www.eon.be, 18.09.2013; interview with the company (and other mentioned sources) 50 see IMJV (2013): http://imjv.milieuinfo.be; 03.12.2103 41 42

51 52

Knauf Insulation (2011a): www.knaufinsulation.be, 23.11.2011 economie.wallonie (2011): www.economie.wallonie.be, 01.12.2011

163

products themselves should be further improved53. On the corporate level, Knauf Insulation mentions some environmental improvements of its products, which also may have an impact on the PM10 performance of the Visé plant54: 

higher shares of recycled glass in the glass wool production



energy savings through improved process control and production processes



efficient packaging (less transportation necessary)



waste reduction

Lafarge Zement Karsdorfer Zement GmbH The cement plant at Kall/ Sötenich (Regio Aachen) is one of the three cement plants of Lafarage in Germany. The production of clinquer was stopped in 2009, but the grinder plant is still operating and produces appr. 0,45 Mio. tons cement per year, which is done by 20 employees 55. In addition, the company has a new operating permit and therefore theoretically could re-launch the furnace56. The available emission data for the plant is a little bit inconsistent. In the E-PRRT/EPER data, PM10 is only included for 2004 (52,5t) and lies slightly above the reporting benchmark. But a check of the Emissionskataster Luft NRW57 has shown that a quite equal value is reported there for 2008 (52,4t). Normally, this value should have been reported to the E-PRTR, as it was done for other pollutant releases in that year. In addition, the data from the Emissionskataster Luft NRW for 1996/97 (Wülfrather Zementwerk GmbH, 59,6t) indicates that the company had quite constant PM10 emissions for a long time. This implicates that it improved its dust/ PM10 performance very early and/or didn’t do (further) reduction measures since then. O-I Manufacturing O-I Manufacturing Netherlands B.V. (Maastricht) is also known under its former name BSN Glasspack, which was purchased by O-I in 2004. O-I owns a lot of plants in Europe and NL and is specialized in glass packaging for food and beverages 58. A combination of E-PRTR data and the Netherlands PRTR59 delivers a long term overview about BSN Glasspacks/ O-I’s PM10 development. All in all, the PM10 emissions were significantly reduced (appr. 90%) in the last twenty years and it seems, as if this was achieved during different investment periods. The company’s current sustainability policy mainly focuses on life cycle assessments and the improvement of production processes, e.g. higher energy efficiency and a higher use of recycled glass60. As the company isn’t a big pollutant anymore, no more explanations are given here. RWE Power AG (Power Plant Weisweiler) RWE Power AG is one of the biggest energy suppliers in Europe, with appr. 15.000 employees and an operating profit of 2,7 Bn. € in 201161. The explanations here mainly focus on the Power Plant at Weisweiler (Regio Aachen). Another example of air quality activities (Hambach Opencast Mine) is given later in Section 2. RWE’s plant at Weisweiler is a coal fired power plant, which gets its fuels by 53

Knauf Insulation (2011b): Belgiens und gleichzeitig Europas größte Produktionsanlage für Glaswolle entsteht in Visé. www.knaufinsulation.com, 25.11.2011 54 Knauf Insulation (2011b) 55 Lafarage 2012: www.lafarge.de/wps/portal/de/Soetenich, 02.11.2012 56 Kölner Stadt Anzeiger (2012): 15 Mann halten Betrieb aufrecht, 12.11.2012 57 LANUV (2013): www.lanuv.nrw.de/luft/emissionen.htm, 03.12.2013 58 BSN Glasspack (2011): www.bsnglasspack.com, 12.09.2011 59 see RIVM (2013): www.emissieregistratie.nl/erpubliek/erpub/facility.aspx, 04.12.2013 60 BSN Glasspack (2012): www.bsnglasspack.com/Sustainability/Sustainability-Goals/, 12.12.2012 61 RWE Power AG (2012c): www.rwe.com/web/cms/en/59722/rwe-power-ag/our-company/, 13.12.2012

164

the nearly located opencast mine Inden62. The net capacity of the plant is appr. 2.600MW, which is achieved by 6 coal fired units that operate in the base load. Since 2007/2008 two of these units are supported by topping-gas turbines that can enable an operation in the intermediate and peak load 63. As it can be expected for a large base-load power plant, the KW Weisweiler is a constant - and one of the biggest - PM10 sources in the EMR. Nevertheless, there are some significant fluctuations in the PM10 levels from one year to another. The biggest range (142t) can be seen from 2007 (374t) to 2008 (516t), where also the minimum and maximum PM10 level of the plant can be observed in the E-PRTR data. Consider: This range is almost as much as 3 “small” point sources which are close to the E-PRTR benchmark (50t/a). Besides installations for flue-gas desulphurization (FGD) and denoxing, which were already installed in 1989, the plant uses of course also de-dusting technologies.

62

RWE Power AG (2012d): www.rwe.com/web/cms/en/60026/rwe-power-ag/locations/lignite/inden/, 13.12.2012 RWEPower AG (2012a): www.rwe.com/web/cms/de/60142/rwe-power-ag/standorte/braunkohle/kw-weisweiler/; REW Power AG (2012b): http://www.rwe.com/web/cms/en/60142/rwe-power-ag/locations/lignite/weisweiler-power-plant/, 06.12.2012 63

165

3.8.4. PM10 Concentrations and Air Quality Planning in the EMR Because the limit values in the European Ambient Air Quality Directive (2008/50/EC) and the corresponding national legislation refer to PM10 concentrations (not directly to emissions), the analysis of concentration values and the strategies for air quality improvement – mainly air quality and action plans – is combined in this section. A strong connection to other activities of the PM-Lab project is also obvious. Similar to the description of the PM10 sources, it is necessary to give an overview about the PM10 concentration “hotspots” in the EMR first. Several PM10 measurement stations are located in the EMR, but not all of them show exceedings of PM10 limit values. In this section, only those areas are taken into account, where at least one limit value exceeding occurred or that had other PM10 related problems (e.g. heavy metal components in PM10 at Genk). Areas which have a good air quality (often mainly belonging to the regional background) are not considered. The second part of section 2 presents the results of case studies, this time focussing on air quality planning in the Euregio Meuse Rhine (EMR). 3.8.4.1. PM10 Concentrations in the EMR For a quick overview of PM10 hotspots in the EMR, a simple indicator has been derived. The PM concentration values from 2007 to 2011 64 were taken and only those areas and stations were selected, which had at least one limit value exceeding in this period, assuming that these areas are relevant in relation to the EU ambient air quality directive. In addition, the cities Genk (BE) and Roermond (NL) were added to the overview. At Genk there are no exceedings of the total PM10 limit values, but it has been taken into account because of problems regarding heavy metal concentrations in PM10. Although there is no measurement station located in Roermond, the city had to implement an air quality plan in 2006 (see below) and is therefore considered in the analysis. Furthermore, there may be other air quality related activities, which cannot be presented here. For example an air quality plan is in preparation at Düren (DE), because NO2 limit values were exceeded. The mentioned indicator, on which the map below (Figure ) is based, is simply the ratio of years with daily PM10 limit value (50µg/m³) exceedings and years with measured PM10 values. The daily limit value is chosen, because the annual limit value for PM10 isn’t exceeded at any measuring station, even at the further mentioned “hotspots”. Doing so, the following spatial pattern occurs between 2007 and 2011: Aachen, Maastricht and the area around the Hambach Opencast Mine show two years with exceedings; Hasselt, Heerlen and Sittard-Geleen only one year (20%). For all of these areas PM10 values are available for all 5 years. Interestingly, in 2011 values far above those in the years before occurred at the Provincie Limburg’s (NL) measuring stations at Maastricht and Sittard-Geleen. For Maastricht this can be mainly explained by the movement of the station from the roof of the Provincial administration next to the A2 highway and the construction activities near the latter. The “hotspot” stations that show the most exceedings in the EMR are all located in the Province Liège: Engis, Seraing-Jemeppe and Saint-Nicolas show 3 exceedings in 5 measured years. Interestingly, the years with exceedings were exactly the same at Seraing-Jemeppe and Engis. The stations at Herstal and Liège (ISseP) show exceedings in quite every year since they started measuring in 2008 and 2009, respectively. As mentioned before, Genk (no exceedings) and Roermond (no station) are special cases, which are also included in the analysis. Further details about PM10 measurements can be seen in the reports of other actions of the PM-Lab project.

64

Note: some plans had to be imlpemented because of earlier exceedings (e.g. AP Hambach, LKP Maastricht)

166

Figure 6. PM10 limit value exceedings for and local air quality and action plans in the EMR Source: own draft (further sources see figure)

3.8.4.2. Case Studies - Air Quality Planning in the EMR 3.8.4.2.1. Netherlands National and Regional Framework65 The regional air quality activities in the Provincie Limburg (NL) are deeply integrated into the national framework of the “National Air Quality Cooperation Programme” (Nationaal Samenwerkingsprogramma Luchtkwaliteit; NSL). The NSL is part of the Dutch air quality law (“Wet Luchtkwaliteit”) and concentrates several air quality related programmes and measures. According to the NSL, the Dutch national state, the provinces and the municipalities together develop reduction programmes, if air quality limit value exceedings occur. Therefore there is a corresponding regional programme in the Provincie Limburg, the “Limburgs Samenwerkingsprogramma Luchtkwaliteit” (LSL), too. The LSL is an official part of the NSL and contains a regional programme (“Regionaal Programma Luchtkwaliteit”; RPL) and a cooperation platform (platform lucht limburg).66 Funding of the LSL activities comes especially in the initial phases from the NSL programme and can be raised later, if further measures are necessary67.

The main air pollution sources in focus of the LSL are large spatial planning projects (‘in betekenende mate (IBM) projecten’; e.g. construction of industrial areas, residential housing, office buildings or infrastructure) 68, road traffic69 in cites and on the superior road network and livestock 65

this part is mainly based on: Provincie Limburg (2008): Limburgs Samenwerkingsprogramma Luchtkwaliteit p. 5 67 p. 33 68 p. 14 66

167

production70. Regarding the latter, the focus especially lies on poultry farms in agricultural reconstruction areas, but no farm with limit value exceedings is located in the EMR (details see chapter 6 of the LSL). Traffic hotspots are identified with the help of the so called “sanergingstool”71. The calculations and projections of air quality values that are made with this tool consider reduction measures as well as new sources. Even if the effects of other measures are included in the calculations, some hotspots can remain and define an additional reduction task (“saneringsopgave”) for the public authorities, for example the A2 in Maastricht, where a tunnel is under construction 72.

Measures which should be implemented by the provincial administration are environmental standards in public transport concessions, implementation of a new provincial traffic/ mobility plan as well as public vehicle procurement and the stimulation of alternative fuel use 73. Beside common regional activities (guidelines for air quality sensitive planning, action plans for hotspots, projects (environmental friendly fuels, environmental zones, greening measures, communication strategy) 74 the local air quality plans75 of the six obliged cities and municipalities play a central role in the RPL. Four of them are located in the EMR (the other are Venlo and Weert) and described now. Local Examples 1. Heerlen The Luchtkwaliteitsplan (LKP) Heerlen76 was implemented in 2007. This plan was necessary, because 19km of streets in the city were above the legislation norm. Therefore, three measure packages were developed, from which 28 measures should be implemented (package II) or further investigated (package III). In 2008 a LKP status report was published, which had a high impact on the on-going air quality planning process in Heerlen 77. Because of changes in the calculation methods in the “saneringstool” and also changes in other calculation factors, a lot of hotspots just disappeared “on the paper” and/ or changed to probable exceeding locations. Therefore a revision of the LKP was necessary and the “Actieplan Luchtkwaliteit Heerlen” (AP) came into charge in 2009. The Actieplan Luchtkwaliteit is still oriented to the targets and the approach of the Luchtkwaliteitsplan, but also considers (new) advices that come from the provincial level (e.g. Limburgs Programma Lucht; LPL). Measures focus on motorized road traffic, bike traffic, motivation of companies, greening, city infrastructure and sustainable planning as well as accompanying measures like communication, city as good example, policy measures and synergies with other projects focusing on mobility, climate change or renewable energy 78. 2. Maastricht In Maastricht the number of hotspots was reduced from eleven to zero between 2005 and 200979. It’s important to state here that there may be similar effects because of the changes in calculation methods and inputs like in Heerlen. So it cannot clearly be said, if the shown developments are results implemented measures of the Luchtkwaliteitplan or changed calculation methods. Nevertheless, it can be assumed that the measures had positive impacts on the PM10 situation at Maastricht. The activities in Maastricht seem also to be shaped by the cooperation and knowledge

69

p. 21 p.27 71 p. 10/ p. 21 72 see A2 Maastricht: www.a2maastricht.nl, 06.12.2013 73 Provincie Limburg (2008), p. 17 74 p.13 75 p. 13/ p. 18 76 Gemeente Heerlen (2007): Luchtkwaliteitsplan – Eindconcept; Gemeente Heerlen: Samenvatting Luchtkwaliteitsplan gemeente Heerlen (LKP Heerlen) 77 Gemeente Heerlen (2009): Actieplan Luchtkwaliteit Heerlen 78 p. 3 79 Gemeente Maastricht (2010): Eindevaluatie Luchtkwaliteitplan Maastricht 2010, p.15 70

168

exchange on the provincial level. They focus on different measures80 in the traffic sector, buildings, green spaces and corresponding activities like communication, monitoring and evaluation. Most measures in the LKP were implemented until 2010 81. Exceptions are the green measures at the Prins Bischopsingel and the traffic dosing light at the Willem Alexanderweg. For these measures alternatives have to be developed. Furthermore, the implementation of environment orientated parking fees and a congestion charge wasn’t possible, because of uncertainties about the effects and the existing legislation. The city’s activities for air quality improvement will continue after the realisation of the LKP measures. For example the mobility information portal ww.maastrichtbereikbaar.nl will be continued, as well as air quality measurements (for NSL reports) and the cooperation in the Platform Lucht. Some suggestions from the Platform Lucht serve as a basis for new measures, which are planned or prepared for decision making. 3. Sittard-Geleen In Sittard-Geleen the Luchtkwaliteitsplan (LKP)82 was developed after the air quality report 2004 identified several PM10 hotspots. The plan was finally implemented in 2007. A forecast for 2010, which is also included in the LKP, showed no more PM10 hotspots, but still some for NO2. Nevertheless, the daily limit value for PM10 was exceeded at the measuring station Asterstraat in 2011, but so far this has no further consequences, because the station is ‘only’ part of the Provincie Limburg’s (NL) voluntary measuring network and doesn’t belong to the Landelijk Meetnet Luchtkwaliteit (LML)83, which is relevant for reporting to the EU commission. Regarding the measures84 in the LKP, a strong focus seems to lie on the connection of the cities accessibility with effects on air quality, e.g. synergies and complementarities with regional activities like the accessibility strategy Tripool-Region and the network analysis Zuid-Limburg. Connections with the LKP can be found in the cities’ susatainability plan 2008-201685, where especially the part „sustainable mobility” refers to the LKP. 4. Roermond Another city in the EMR that had to implement an air quality plan is Roermond86, because the PM10 calculations for the air quality report 2005 showed that some hotspots will occur in 2004/2005. The air quality plan mentions as measures a checklist for spatial development projects, measures that are already included in other (policy) documents and at last traffic measures, which had already started before or were implemented in 2005. In addition, the need for a measuring station at the hotspot Keulsebaan was formulated, but this station doesn’t exist until today. Positive effects on the city centre are especially expected through the construction of the A73. Similar to Sittard-Geleen, connections to the LKP are made in the environmental policy plan2008-201187.

80

p.18ff p.36 82 Gemeente Sittard-Geleen (2007): Luchtkwaliteitsplan Sittard- Geleen 83 see RIVM (2013b): www.lml.rivm.nl, 03.12.2013 84 Gemeente Sittard-Geleen (2007), p.23ff 85 Gemeente Sittard-Geleen: Duurzaahmheidsplan 2008-2016 86 Gemeente Roermond (2006): Luchtkwaliteitplan Roermond 2004/2005 - Actualisatie luchtkwaliteit, inventarisatie van knelpunten en maatregelen 87 Gemeente Roermond (2008): Milieubeleidsplan gemeente Roermond 2008 - 2011 81

169

3.8.4.2.2. Germany National and Regional Framework The national air quality strategy of Germany is based on four main lines: definition of air quality standards, emission reduction requirements according to the best available technologies, product regulations and definition of emission ceilings88. Important instruments on the national level are the Federal Immission Control Act (Bundes-Immissionsschutzgesetz; BImSchG) and the corresponding implementing ordinances (e.g. Technical Instructions on Air Quality Control (TA Luft), Amendment to Ordinance on Small Firing Installations (1. BImSchV), Implementation of the EU Industrial Emission Directive, Transboundary air pollution control policy). The European ambient air quality directive (2008/50/EC) is mainly transferred into German law by combination of the above mentioned Federal Immission Control Act and the implementing ordinance “Verordnung über Luftqualitätsstandards und Emissionshöchstmengen (39. BImSchV)”89. The execution of the BImSchG and the ordinances are tasks of the federal states like North Rhine-Westphalia (NRW)90 and therefore the responsibility for the implementation of air quality plans, too. In the federal state NRW this task is taken over by the district authorities (Bezirksregierungen). For the German part of the EMR the Bezirksregierung Köln is the responsible authority 91 that for example is author of the air quality plans at Aachen and Hambach.

Local Examples 1. Aachen The integrated air quality and action plan for the city of Aachen92 had originally to be implemented, because of a limit value exceeding for NO2 in 2006. But in 2007 also the daily limit value for PM10 was exceeded, so that an action plan had to be developed additionally93. The limit value for PM10 was exceeded again in 2009, while the values in 2010 (34) and 2011 (34) were very close to it. The plan integrates three different approaches: 

air quality plan



action plan



environmental zone

As usual, most measures of the air quality plan94 have a medium to long term focus and are updated continuously. The city itself categorizes the measures into “mobility” (28 measures; mainly promotion of environmental friendly means of traffic) and “energy” (5 measures). The main focus lies on the first, while the latter are regarded to have complementary positive effects on the air quality in Aachen. The action plan95 focuses on the traffic hotspot Wilhelmstraße, where also the equally 88

BMU (2013a): www.bmu.de/themen/luft-laerm-verkehr/luftreinhaltung/kurzinfo/, 04.12.2013 BMU (2013b): www.bmu.de/service/publikationen/downloads/details/artikel/verordnung-zur-durchfuehrung-des-bundesimmissionsschutzgesetzes/, 04.12.2013 90 MKULNV NRW (2013): www.umwelt.nrw.de/umwelt/immissionsschutz/immissionsschutzgesetz/index.php. Ministerium für Klimaschutz, Umwelt, Landwirtschaft, Natur- und Verbraucherschutz (MKULNV) des Landes Nordrhein-Westfalen, 04.12.2013 91 Bezirksregierung Köln (2013): www.bezreg-koeln.nrw.de/brk_internet/organisation/abteilung05/dezernat_53/plaene/index.html, 04.12.2013 92 Bezirksregierung Köln (2009): Integrierter Luftreinhalte- und Aktionsplan der Bezirksregierung Köln für das Stadtgebiet Aachen vom 01.01.2009 93 p. 13 94 p.58ff 95 p.100ff 89

170

named measuring station is located, and includes intensified parking area surveillance, implementation of delivery spaces and the use of most efficient vehicles on the passing bus lines. The main measure of the action plan is the temporary closure of the Wilhelmstraße in the southern driving direction (i.e. road side, where the measuring station is located). This measure was implemented in April 2010 and refers to heavy vehicles above 3,5t, which since then aren’t allowed to pass the road between 7 to 12h anymore. In the long term, a freight traffic routing plan was desired and included in the air quality plan (measure M12), but was skipped later on 96. The environmental zone97 serves as an additional option and should only come into action, if all other measures will not achieve the desired air quality improvements. It may cover the city centre and the most populated surrounding areas. Positive effects are mainly expected by the reduction of freight traffic. 2. Hambach Opencast Mine (Rhenish Lignite Mining Area) Since 2005 an action plan98 was in charge around the Hambach Opencast mine. This plan came into action, because the daily limit value was exceeded in 2004 (in 2006 35 exceeding day’s). The plan mainly obliged the owner company of the opencast mine (main local PM10 source), RWE Power AG, to implement reduction measures. Because the limit value was exceeded again in 2010 and 2011, an air quality plan99 had to be developed, which came into charge by the end of 2012. Besides other information, the air quality plan includes different technical and organisational measures of the RWE Power AG that were already part of the action plan 2006 as well as additional measures that the company implemented (more or less) voluntarily until 2011, and finally new measures, which were developed especially for the air quality plan 100. All measures mainly focus on dust binding through water spraying and dust prevention by cleaning or other organisational improvements. In addition to the action plan, the air quality plan includes also some measures of the surrounding cities and municipalities (Bergheim, Elsdorf , Kerpen, Niederzier), but these are mainly existing actions in the traffic and energy sector, which aren’t specifically developed for the AQP 101. This is also the case for further mentioned options102. 3.8.4.2.3. Belgium National and Regional Framework In Belgium the competencies in the field ‘environment and health’ are divided between the federal state, the communities and the regions, but of course other actors like cities, research institutions or citizens play also an important role103. Therefore the national action plan “NEHAP” was developed to coordinate the activities of the responsible authorities. Environmental responsibilities like air quality protection (emission inventories, monitoring, target and standard setting, planning, measures, permits and enforcement for stationary sources, taxes on vehicles) 104 belong mostly to the regions, while the federal state takes over tasks in product regulation, nuclear protection and waste transportation105. Nevertheless, there remain mixed competencies and a need for cooperation at least between the regions and the federal state regarding air quality (e.g. climate protection or air quality surveillance)106.

96

see Stadt Aachen (2011): Integrierter Luftreinhalte- und Aktionsplan der Bezirksregierung Köln für das Stadtgebiet Aachen vom 01.01.2009 - aktualisierte Maßnahmenübersicht letzter Stand: 15.02.2011 97 p. 95 98 Bezirksregierung Köln (2005): Aktionsplan in der Umgebung des Tagebaus Hambach; BR Köln, 29. September 2005 99 Bezirksregierung Köln (2012): Luftreinhalteplan Hambach vom 31.12.2012 100 p. 71ff 101 p. 52ff 102 see p. 103 103 NEHAP (2012): www.ogm-ggo.be/eportal/Aboutus/relatedinstitutions/NEHAP. FOD Volksgezondheid, Veiligheid van de Voedselketen en Leefmilieu, 26.12.2012 104 Nieuwejaers, B.: Air pollution policy: Institutional and administrative set-up. Presentation. LNE Vlaanderen (presentation), p. 4 105 NEHAP 2012 (see above) 106 AWAC (2007): Plan Air Climat, p.17

171

Walloon Region The main document that gives an insight in the recent air quality strategy of the Walloon Region is the so called “Plan Air-Climat”107. This plan can be seen as a kind of general strategy or vision for air quality improvement and climate protection. The plan includes nearly 100 measures, which are grouped into 9 domains. From these 100 measures, at least 24 have a more or less strong connection to PM10108. It’s important to state that a lot of these measures focus on the foundation of organisational structures as basis for further air quality improvements, e.g.: o

adaption and controlling of air quality legislation (M2 and M3)

o

foundation of the Walloon Agency for Air and Climate Change (AwAC) (M4)

o

extension of the air quality measuring network (M5)

o

development of an alert system for pollution peaks (M18)

The plan itself promises a lot of improvements and for example all the four above mentioned measures were implemented. But nevertheless, air quality or action plans in the sense of the European Ambient Air Quality Directive (2008/50/EC) still lack on a local level in Wallonia and so in the Province Liège, too. One already implemented measure (No. 18) from the Plan Air-Climat, which comes close to such a plan, is the so called “Plan Pics de Pollution”109. The main purpose of this “sub-plan” is the set-up of an alert system for pollution peaks, including the development of an information and action procedure, which are dependent on certain peak levels. All the procedures involve different actors on different levels. This includes also the cities and municipalities in the Province Liège. Especially the activities (several prohibitions and orders) which have to be done on the local level in the zones Engis and Liège are interesting for the PM-Lab project. All in all the included measures in the Plan Pics de Pollution have only a really short term and rather reactive focus110. On the one side, an advantage of the plan is that it serves as a kind of interface between different actors and administrative levels (also between the Walloon and Flemish Region). On the other side, one may ask, whether the implementation of this plan can serve as an explanation for the lack of formal local air quality plans in Belgium. The (necessary) implementation of the plan may have “locked” resources of personnel, finance and know how that could have been used for the development of a long term strategy (note: other measures of the Plan Air-Climat have a more long term focus). In addition the cities’ participation in this plan could deliver a feeling of “doing something”, so that further or other measures aren’t taken into account. But, as will be shown in the next section, some cities like Liège are aware of air quality problems and do some actions, independently of the existence of a formal air quality plan. Local Examples Walloon Region/ Prov. Liège: 1. Liège Agglomeration As shown above, a lot of areas in the Province de Liège face problems with high PM10 concentrations. Therefore all the cities and municipalities with such problems were contacted (in English and French) by the PM-Lab project. Unfortunately, only the cities Liège and Herstal reacted and gave the possibility for a personal interview. Therefore it is really hard to evaluate the efforts for air quality improvement of the other cities. In fact, none of the cities in the Liège agglomeration 107

AWAC (2007): Plan Air Climat Interview with AWAC 109 AWAC (2009): Plan d’actions en cas de pic de pollution par les poussières fines 110 AWAC (2009): Plan d’actions en cas de pic de pollution par les poussières fines, p. 3; Interview with City of Liège 108

172

implemented a formal air quality or action plan in the sense of the European Ambient Air Quality Directive (2008/50/EC). So the question whether there is done more than the “business as usual” (e.g. environmental permits for companies, actions from the Plan Pics de Pollution) has to be left open. 2. Liège City The city of Liège has also no formal air quality or action plan, but there are a lot of activities, which theoretically could be part of such a plan, e.g.



Plan d‘Action Pics du Pollution (see above)



Projet de Ville 2007-2015111



Plan Communal de Mobilité 2004112

Some actions, mainly in the traffic sector were already taken in the former Plan de Ville 2003, e.g. improved car parking and several actions for the promotion of environmental friendly means of traffic113. The current Projet de Ville 2007-2015 continues this approach and includes at least 15 new actions that could be part of a formal air quality plan, especially the planning of a tramway, one of the “grandes projets”, can be seen as a serious attempt114. Flemish Region The Flemish region has set up a broad approach for the improvement of the regional air quality. A lot of measures were taken on different levels at least since 2005, when the “Vlaamse Stofplan”115 was published. Additional measures were included in the Luchtkwaliteitsplan (LKP) 2008 116 - a kind of update of the Vlaamse Stofplan - and other plans. The actions of the Flemish Region focus on the following sectors: traffic, industry, shipping, households and tertiary sector, cities and municipalities. In addition to this sectoral focus, also a spatial differentiation has taken place, so that some measures focus on the whole Flemish Region, while others are only relevant for specific locations (i.e. hotspots) or the different Flemish reporting zones. In addition to the Vlaamse Stofplan, the LKP 2008 underlines a strong and explicit connection to regional NEC-policy (reduction of secondary PM10) and the regions’ climate change plan. According to the LKP, the activities of the Flemish region also stimulated further air quality initiatives and plans117. Local Examples Flemish Region/ Prov. Limburg (BE): 1. Genk Today, Genk is the only city in the Belgian Provincie Limburg (BE) with PM10 problems 118. As mentioned above, limit values for total PM10 aren’t exceeded at Genk, but the city faces other environmental issues, which are mainly related to the industrial activities at Genk-Zuid119:  

odour PCB

111

Ville de Liège (2008a): Projet de Ville 2007-2015 Ville de Liège (2004): Plan Communal de Mobilité 2004 (Version définitive). Liège, février 2004 113 see Ville de Liège (2008a), p. 4 and Ville de Liège (2004) 114 see Ville de Liège (2008b): Projet de Ville 2007-2015 - Actions prioritaires 115 Vlaams Gewest (2005): Saneringsplan fijn stof voor de zones met overschrijding in 2003 en aanpak fijn stofproblematiek in Vlaanderen. 116 Vlaams Gewest (2008): Luchtkwaliteitsplan en uitstelaanvraag voor de normen van PM10. 117 Vlaams Gewest (2008), p.18 118 Interviews with Provincie Limburg and City of Genk 119 this section is mainly based on Stad Genk (2012): Actieplan Genk-Zuid/ E-missie plan Genk-Zuid versie juni 2012 112

173

 

dioxins heavy metal concentrations in PM10 (see Action 5 of the PM-Lab project!)

For the PM-Lab project especially the heavy metal concentrations in PM10 are interesting. While Action 5 intensively focuses on chemical analyses, this section delivers complementary information about the strategy that is used to face the environmental and especially the air quality problems120. The problems around Genk-Zuid came first on the agenda in 2005, when it got evident that the Nickel concentrations in fine dust were too high. As one main source the company APERAM (formerly Ugine & Alz/ ArcelorMittal Genk) came into suspicion (directed as well as diffuse emissions), but also other companies are responsible for emissions at Genk-Zuid. Since 2006 a lot of measuring campaigns have been done and also diffuse sources in the production processes came into focus.

But measuring alone doesn’t solve air quality problems; it can only serve as basis for the measure development and evaluation. Therefore some measures were implemented by the city of Genk (e.g. moving of a public school) as well as by the companies, especially APERAM (e.g. adaption of operation and production processes, coverage of slag boxes.) that will invest in further measures in 2012121. One landmark of the activities at Genk-Zuid is the long lasting and intensive cooperation in the local taskforce “Stuurgroep Genk-Zuid”. Partners are the Flemish Government (i.e. LNE), Provincie Limburg, City of Genk, companies and several health and research institutions (e.g. VMM). The tasks122 are shared corresponding to each partners’ core competencies. For example, the Provincie Limburg (BE) takes over the organisation of the task force, while the VMM contributes with measurements and the so called “human biomonitoring”. The city of Genk uses its proximity to the citizens and therefore focuses on information of consultation and communication with the public.

Regarding the improvement of air quality, the e-missie plan 2012 includes 56 measures, with two main lines123: 1. emission reduction (33 measures: i) general, ii) air, noise, soil, water) 2. healthy lifestyle (23: measures: i) health, ii) communication)

120

see Stad Genk (2012), p. 4ff p. 6/ Interview 122 p. 13ff 123 p. 29ff 121

174

3.8.5. Summary and Conclusion Action 4.2 The Euregio Meuse-Rhine shows different emission and concentration situations. One the one side, there are (only) traffic affected cities, while on the other side especially cities in the Province Liège are additionally affected by industrial sources. Some situations like at Genk (heavy metal concentrations) or around the Hambach Opencast mine are quite unique. These different situations also mirror in different ways of action. Local activities in Dutch and German parts of the EMR are mainly directed to traffic, which is the main source there. But also improvements in the industrial sector, often guided by superior levels, can be observed in the whole EMR. This can be seen for example in the rather positive development of E-PRTR point sources.

Measure implementation seems to be a mix of legislation and cooperation and is done this way in most parts of the EMR. Here the question rises, if it is a voluntary cooperation or „cooperation by legislation”. Cooperation seems to be most intensive and systematic in the Netherlands, but also other parts of the EMR have different air quality work groups, e.g. the Stuurgroep Genk or work groups around the air quality plans in Germany. It has been shown that the cities in the Province Liège lack formal air quality or action plans so far, while these are implemented in other EMR parts. The question is, whether this can (partly) be explained by a lack of cooperation. If yes, this can also be seen as a chance: When combined with a regional knowledge exchange in the EMR and beyond, such a “cooperative air quality planning” model in the agglomeration or province Liège promises great chances for air quality improvements in one of the EMR’s most populated and also most polluted parts.

But there is also much potential for more cross-border cooperation in the Euregio Meuse-Rhine. As the analysis has shown, activities until now mostly take place within the national frameworks. Cooperation should start with rather ‘soft’ measures (low costs, easily transferable) like knowledge exchange about legislative and technical developments that can lead to collaborative learning processes. It should take place in an Euregional air quality network, where especially local and regional policy makers take part and not only for example within a rather scientific community. One possible ‘hard’ outcome of such cooperation, for example, could be a common air quality report, may it be an unofficial or official (report to the EU commission) document. Such a report can build the basis for a cross-border discussion about common and targeted measures which may be included in an Euregional air quality plan that includes specific (city, region) as well as common (Euregio) measures. The outcomes of the PM-Lab project give a jump start for this, but the implementation depends on the willingness of decision makers in the Euregio Meuse-Rhine.

175

4. Speciation – Source recognition – health effects

4.1. Introduction The exposure to metals in suspended particles (particulate matter, PM) is often above natural background levels due to anthropogenic processes. Enhanced metal concentrations can pose serious risk to human health since they can be absorbed into human lung tissues during breathing. A great deal of research has focused on the metal composition of atmospheric suspended PM. However, determination of metal levels is usually limited to total metal concentration. Although measurements of the total metal concentration can give an overall indication of the general pollution level, they provide no information on the chemical speciation of the metals and thus the real toxicity. Therefore, it is important to quantify dedicated chemical metallic forms (species) since bioavailability, solubility, geochemical transport largely depend on physical-chemical speciation. The knowledge of the chemical speciation is vital in understanding the effects on human health. Currently, the most widely used method for characterization of the chemical speciation is particle sampling on filters. Depending on the kind of analysis, the type of filter is selected, bringing into account the capacity of the sampling devices. The locations are selected in relationship with sources present near this locations. In combination with other available analytical data (VMM, VITO, Lanuv, ISSeP, RIVM), results can be interpreted. Within this project the speciation of Chromium (Cr), Nickel (Ni) and Platinum (Pt) was studied: Chromium (Cr) Chromium has two predominant oxidation states in the atmosphere: Cr(III), which is an essential nutrient in low doses and Cr(VI), which is highly toxic and carcinogenic. Chromium is used in the tanning, dyeing and plating industries. Nickel (Ni) Nickel is well known as a human carcinogen. Occupational exposure has been found to correlate with an increase in the frequencies of prostate, skin, oesophageal, nasal and lung cancers in particular. All Ni compounds are classified as potential human carcinogens, they are divided into a number of groups with different intrinsic cancer risks. The specified categories identified are water-soluble, sulfidic, metallic and oxidic nickel. Exposure to nickel oxides and sulfides, which have low solubility in water, has been recognized as one of the prominent causes for occupational nickel-related lung and nasal cancer. Ni sulfides are the most carcinogenic Ni compounds. Metallic nickel is known to provoke pronounced allergic reactions. The carcinogenic potential of water-soluble Ni compounds continues to be discussed. Platinum (Pt) The platinum group elements (PGEs), particularly platinum, palladium and rhodium are increasingly emitted into the environment from automotive catalytic converters which employ these metals as exhaust catalysts. Recent studies on PGEs toxicity and environmental bioavailability indicate that environmental exposures to these metals may pose a health risk, especially at a chronic, subclinical level.

176

4.2. Experimental approach Chromium (Cr), Nickel (Ni) and Platinum (Pt) are selected as target elements for different sources (industrial activities and traffic). Sampling locations are therefore selected in Genk, Maastricht and Herstal using dedicated sampling procedures. 4.2.1. Chromium (Cr) The sampling location was situated at Genk and Herstal, nearby an industrial zone. Fine dust was collected on an alkaline impregnated ashless cellulose filter. After extraction of the filter in an alkaline medium, Cr(VI) was separated from Cr(III) by means of IC (ion chromatography). The analyses of Cr(VI) was performed by an on-line coupling with Inductively Coupled Plasma – Mass Spectrometry (ICP-MS). 4.2.2. Nickel (Ni) The sampling location was situated at Genk and Herstal, nearby an industrial zone. After sequential extraction of particulate matter (separating soluble, sulfidic, metallic and oxidic nickel), Ni (=341 nm) analyses in the extracts were performed with Inductively Coupled Plasma – Optical Emission Spectrometry (ICP-OES). 4.2.3. Platinum (Pt) The sampling location was situated in Maastricht, nearby heavy traffic. After digestion of the filters in strong acids, Pt was analysed by means of ICP-MS.

177

4.3. Results 4.3.1. Cr speciation 4.3.1.1. Method The experimental approach used is based on the methodology described in the article: Determination of hexavalent chromium in ambient air: A story of method induced Cr(III) oxidation Kristof Tirez et al. Atmospheric Environment 45 (2011) 5332-5341 Fine dust is collected on an alkaline impregnated ashless cellulose filter (Whatman 40) during 23 hours. After extraction of the filter in an alkaline medium, Cr(VI) is separated from Cr(III) by means of IC. The detection of Cr(VI) is performed by an on-line coupling with ICP-MS. Sampling Sampling was performed on weekdays. The samples were recovered immediately after sampling and transported to the laboratory in a cooled transport ( < 5C). 4.Sampling time Two filters: each 11 h (Flow: 2.3 m3/h) Filters Ashless cellulose filters: Whatman 41 (47 mm – 20 m) In order to evaluate concentrations and conversions between Cr species, 4 separate samplings were performed simultaneously, the filters are impregnated with 0.12 M NaHCO 3 : 2 duplo filters 1 filter spiked with 20 ng Cr(VI) (equivalent to 0.39 ng Cr(VI)/m³) 1 filter spiked with 1000 ng Cr(III) (equivalent to 19 ng Cr(III)/m³) Sampling location - Genk (GK11), this location is situated at 190 m of a stainless steel factory in the predominantly downward wind direction. - Herstal, nearby an industrial zone Sampling period Genk: - 10/10/2011 until 21/12/2011 - 02/05/2012 until 05/07/2012 Herstal: - 5/03/2012 until 26/04/2012

178

4.3.1.2. Results Table 1. Summary of the Cr(VI) measurements for Genk (2011-2012) and Herstal (2012)

Mean Cr(VI) concentration (ng/m3) Range Cr(VI) concentration (ng/m3) Average recovery Cr(VI) on spiked filters (%) Average oxidation of Cr(III) on spiked filters (%) Cr(VI)/Cr ratio (%)

Genk 2011 October-December 3,1 < 0,05** - 20 52 ± 33 1,5 ± 1,0 3,3

Genk 2012 May-July 0,95 < 0,02* – 6,18 58 ± 22 1,3 ± 0,8 2,0

Herstal 2012 March-April 0,11 < 0,02* – 0,28 64 ± 21 1,0 ± 0,5 1,1

** sampling volume = ± 30 m³ * sampling volume = ± 50 m³

The mean concentration of Cr(VI) for Genk (2011-2012) is on average above the air quality guideline value of 0.20 ng Cr(VI)/m³ in PM10, recommended as an annual average by DEFRA (DEFRA = Department for Environment, Food and Rural Affairs ). The mean concentration of Cr(VI) for Herstal is below this air quality guideline. The mean concentration of Cr(VI) for Genk (2011) is above the health standard of 2.5 ng/m³ according to the RIVM. The mean concentration of Cr(VI) for Genk (2012) and Herstal (2012) is below this health standard. Cr(VI) concentrations Genk (2011-2012) – Herstal (2012) Figures 1-2-3. Cr(VI) concentrations (ng/m³) measured in Genk (2011-2012) and Herstal (2012). (Appendix, p 15-16) Figures 4-5-6. Average Cr(VI) concentrations (ng/m³) with the standard deviation. (Appendix, p 17-18) Conversions between Cr species Tables 2-3-4. Conversions between Cr species for Genk (2011-2012) and Herstal (2012). (Appendix, p 19-20-21) Figures 7-8-9. The average Cr(VI) concentration and the concentration of Cr(VI) on the spiked filters with Cr(VI) and Cr(III). (Appendix, p 22-23) Tables 5-6-7. Cr(VI)/Cr ratio (%) for Genk (2011-2012) and Herstal (2012). (Appendix, p 24-25-26) Figures 10-11-12. Cr(VI) concentration compared with the total Cr concentration. (Appendix, p 27-28)

179

The uncertainty on the measurement is defined by the oxidation of Cr(III) to Cr(VI) and the reduction of Cr(VI) to Cr(III). Table 8. Cr(VI) concentrations after corrections for oxidation and reduction for Genk (2011-2012) and Herstal (2012).

ng Cr(VI)/m³

Genk 2011 October-December Without With correction correction oxidation oxidation 3,10 1,92

Genk 2012 May-June-July Without With correction correction oxidation oxidation 0,95 0,38

Without correction for Cr(VI) reduction With correction for 5,96 3,69 1,64 Cr(VI) reduction Cr6+measured= (Cr6+actual*freduction) + (Cr3+actual*f oxidation)

0,66

Herstal 2012 March-April Without With correction correction oxidation 0,11 0,03 0,17

0,05

4.3.1.3. Conclusion At Genk, a significant decrease of the Cr(VI) concentration between the period of October-December 2011 and May 2012 is observed. The mean Cr(VI) concentration in 2012 still is above the the air quality guideline value of 0,20 ng Cr(VI)/m³ in PM10. The Cr(VI) concentration measured at Herstal is significant lower than the concentration at Genk. The mean Cr(VI) concentration is lower than the air quality guideline value of 0,20 ng Cr(VI)/m³ in PM10. Genk 2011-2012; Herstal 2012: Contribution of the wind direction to the average Cr concentration depending on the wind speed. (Appendix, p 29-30-31)

180

4.3.2. Ni analyses The experimental approach used, is based on the methodology described in the article: Speciation of Nickel in airborne particulate matter by means of sequential extraction in a micro flow system and determination by graphite furnace atomic absorption spectometry and inductively coupled plasma mass spectrometry Lars Füchtjohann et al. J. Environ. Monit., 2001, 3, 681-687 The filters are divided in four equal parts. Two parts are used for the analyses of the total Ni concentration by microwave digestion and two parts for the speciation of Ni using a sequential extraction methodology.

4.3.2.1. Location : Genk – GK11 Sampling period 9-07-2011 until 3-9-2012 (sampling every fifth day – high volume sampler). Sampling location Genk (GK11), this location is situated at 190 m of a stainless steel factory in the predominantly downward wind direction. Total Ni concentration Table 9. Total Ni concentration in fine dust analysed after microwave digestion, measured by VMM during 2010 and UHasselt during 2011-2012. (Appendix, p 32-33-34) Figures 13-14. Comparison of the fluctuations of the total Ni concentration (µg/filter) to the total amount of fine dust collected on the filter. (Appendix, p 35) Figure 15. Comparison of the fluctuations of the total Ni concentration (ng/m³) measured in Genk 2010 (VMM) and 2011 (UHasselt). (Appendix, p 36) Figures 16-17. Comparison of the total Ni concentration (ng/m³) with the target value for Ni, according to the EU directives = 20 ng/m³. (Appendix, p 37) Table 10. Comparison of the average Ni concentration in 2010-2011-2012.

average Ni concentration ng/m³ extremes average fine dust concentration µg/m³

2010 VMM 27,6 1,2 - 246

2011 UHasselt 19,5 1,7 - 155 19,0

2012 UHasselt 13,2 1,6 – 107 24,8

 The average values in 2010 exceeds the target value (31/12/2012) for Ni according to EU directives: limit value = 20 ng/m3.  The average values in 2011-2012 are below the target value of 20 ng/m3. The average Ni concentration is decreasing from 2010 to 2012. Genk 2011-2012: Contribution of the wind direction to the average Ni concentration depending on the wind speed. (Appendix, p 38)

181

Speciation Ni The total Ni concentration is divided in 4 fractions with different intrinsic cancer risk. All Ni compounds are classified as potential human carcinogens. -

Fraction 1: soluble (Nisoluble; NiCO3) Fraction 2: sulfidic (NiS; Ni2S3) Fraction 3: metallic (Ni) Fraction 4: oxidic (NiO)

 The carcinogenic potential of water-soluble Ni compounds is under discussion.  There is no evidence that exposure to metallic Ni increases the risk for respiratory cancer, but it is known to provoke pronounced allergic reactions.  Exposure to Ni oxides and sulfides, which have low solubility in water, has been recognized as one of the prominent causes for occupational Ni-related lung- and nasal cancer.  Ni sulfides are the most carcinogenic Ni compound. Table 11 .Classification of nickel and nickel compounds with respect to carcinogenicity. (Appendix) Table 12. Sequential extraction procedure.

fraction 1 fraction 2 fraction 3 fraction 4

Leaching solution EDTA H2O2/ ammonium citrate KCuCl3 decomposition in HNO3

Ni compounds soluble (Nisoluble; NiCO3) sulfidic (NiS; Ni2S3) metallic (Ni) oxidic (Ni(NO3)2)

The sequential extraction is performed on those filters with a total Ni concentration after digestion above 40 µg/l (detection limit of Ni with ICP-OES = 5 μg/l). Table 13. Ni concentration (µg/filter) analysed in 4 fractions after sequential extraction. (Appendix, p 40) Figures 18-19. Ni concentration analysed after sequential extraction and total Ni concentration. (Appendix, p 41) Table 14. The average Ni concentration present in the different fractions.

µg/filter Total Ni soluble (Nisoluble; NiCO3) sulfidic (NiS; Ni2S3) metallic (Ni) oxidic (Ni(NO3)2)

GENK 2011 13.5 4,5 (19%) 1,8 (7%) 1,4 (4%) 22 (69%)

2012 9.0 3.7 (21%) 1,6 (9%) 0,9 (5%) 14 (65%)

A decrease in total Ni concentration is observed in 2012 compared with 2011, but the distribution of the concentration in the different fractions is comparable. Table 15. Ni concentration (%) analysed in 4 fractions after sequential extraction. (Appendix, p 42)

4.3.2.2.1. Conclusions: The mean fractions of total Ni were found to contain:

182

20 ± 9 % soluble Ni; 8 ± 3 % sulfidic Ni; 5 ± 2 % metallic Ni and 67 ± 11 % oxidic Ni The highest amount of Ni is obtained in fraction 4: Ni is mainly present in fine dust as an oxidic compound (48-83%). The sulfidic and metallic Ni species are less than 14 % present in fine dust.  62-92 % of the total Ni is present in the insoluble form. The soluble Ni fraction is within the 8-38% concentration range. The analyses prove the importance of the speciation determination of Ni. Total Ni concentration doesn’t say anything about its potential danger. 4.3.2.3 Location: Herstal Sampling period 20-02-2012 untill 10-6-2012 (sampling every fifth day – high volume sampler). Sampling location Herstal, nearby an industrial zone Total Ni concentration Table 16. Total Ni concentration in fine dust analysed after microwave digestion. (Appendix, p 43) Figure 20. Total Ni concentration in fine dust analysed after microwave digestion. (Appendix, p 44)

Table 17. Average Ni concentration in fine dust analysed in Genk (2011-2012) and Herstal (2012)

Average total Ni (ng/m³)

GENK 2011 19,5 (1,7 – 155)

2012 13,2 (1,6 – 107)

HERSTAL 2012 3,8 (1,0 – 12,0)

The total Ni concentrations in Herstal are below the target value (31/12/2012) for Ni according to EU directives: limit value = 20 ng/m3. These concentrations are significantly lower than the measured Ni concentrations at Genk (20112012). Herstal 2012: Contribution of the wind direction to the average Ni concentration depending on the wind speed. (Appendix, p 45) Speciation Ni Due to the low concentration of the total Ni, speciation could only be performed on 5 filters.

183

Table 18. Ni concentration (µg/filter) analysed in 4 fractions after sequential extraction and the total Ni concentration.

20/03/2012 25/03/2012 4/04/2012 21/05/2012 26/05/2012

Ni µg/filter Fractions 1

2

3

6,4 ± 0,4 5,0 ± 0,2 3,7 ± 0,1 8,9 ± 0,4 4,8 ± 0,3

3,1 ± 0,1 2,1 ± 0,1 1,1 ± 0,1 3,2 ± 0,1 1,9 ± 0,1

1,2 ± 0,1 0,6 ± 0,1 0,5 ± 0,2 1,5 ± 0,1 1,3 ± 0,3

Total Ni

4

Sum fractions

17 ± 1 9,6 ± 0,3 5,6 ± 0,1 34 ± 1 28 ± 1

28 17 11 48 36

34 ± 2 21 ± 1 12 ± 1 46 ± 10 33 ± 1

Table 19. Ni concentration (%) analysed in 4 fractions after sequential extraction.

Ni %

20/03/2012 25/03/2012 4/04/2012 21/05/2012 26/05/2012

Fractions 1

2

3

4

23 ± 2 29 ± 1 34 ± 1 19 ± 1 13 ± 1

11 ± 1 12 ± 1 10 ± 1 7±1 5±1

4±1 3±1 4±2 3±1 4±1

62 ± 2 56 ± 2 52 ± 1 71 ± 1 78 ± 2

The mean fractions of total Ni were found to contain: 24 ± 8% soluble Ni; 9 ± 3 % sulfidic Ni; 4 ± 1% metallic Ni and 64 ± 11% oxidic Ni  The major fraction of the total Ni concentration consists of oxidic Ni. Figure 21. Ni concentration analysed after sequential extraction and total Ni. (Appendix, p 46) Table 20. Total Ni (%) and average Ni concentration (%) present in the different fractions.

Total Ni soluble (Nisoluble; NiCO3) sulfidic (NiS; Ni2S3) metallic (Ni) oxidic (Ni(NO3)2)

GENK 2011 0.09 19% 7% 4% 69%

2012 0.06 21% 9% 5% 65%

Herstal 2012 0.02 24% 9% 4% 64%

 The distribution of Ni in the different fractions is comparable for Herstal and Genk.

184

4.3.3. Pt analyses The experimental approach used, is based on the methodology described in the article: Platinum Group Elements: A Challenge for Environmental Analytics A. Dubiella-Jackowska et al. Polish J. of Environ. Stud. Vol. 16, No. 3 (2007), 329-345 The filters are divided in four equal parts. 2 parts are used for the analyses of Pt. A certified reference material (road dust BCR-723: 81.3 ± 2.5 μg Pt/kg) was analysed using the same procedure. Table 21. Total Pt concentration analysed in fine dust. (Appendix, p 47) Figure 22. Pt concentration (pg/m³) and total amount of fine dust (µg/m³). (Appendix, p 48) This graph shows the importancy of the analyses of Pt: the ratio of fine dust and Pt in the first period is much higher than in the second period. Figure 23. Mean Pt concentrations with the standard deviation. (Appendix, p 49)

The results were compared with a study, performed in Germany (2008-2009) and which is described in the article: Platinum group elements (Pt, Pd, Rh) in airborne particulate matter in rural vs. urban areas of Germany: Concentrations and spatial patterns of distribution. Fathi Zereini et al. Science of the Total Environment 416 (2012) 261-268 Table 22. Comparison of the Pt concentration in an urban area in Maastricht – Germany and a rural area in Germany

Average Pt (pg/m³)

Germany Rural area (Feb08- Jul09)*

Urban area (Jul08- Aug09)*

1,9 (0,1 – 19,3)

12,4 (1,2 – 80,9)

Maastricht Urban area (Jul11- Sep12) (UHasselt) 9,3 (2,1 – 21,0)

4.3.3.1. Conclusion The concentrations of Pt measured at Maastricht are situated in the same concentration range as the results obtained in an urban area in Germany, while the concentrations analysed in a rural area are significantly lower. Platinum can be used as target element for estimating the contribution of traffic to fine dust (source recognition). Maastricht 2012: Contribution of the wind direction to the average Pt concentration. (Appendix, p 50) Additional Pt measurements were performed in the period June-July 2013 (Table 22, Figures 24-25) . (Appendix, p 51-52)

185

4.4. General conclusions 4.4.1. Chromium At Genk, a significant decrease of the Cr(VI) concentration between the period of October-December 2011 and May-June 2012 is observed. These results reflects the special efforts taken by the local industry to reduce the emission of fine dust. The mean Cr(VI) concentration in 2012 still is above the air quality guideline value of 0.20 ng Cr(VI)/m³ in PM10. The Cr(VI) concentration measured at Herstal is significant lower than the concentration at Genk. The mean Cr(VI) concentration is lower than the air quality guideline value of 0.20 ng Cr(VI)/m³ in PM10. 4.4.2. Nickel The average Ni values in Genk measured in 2010 exceeds the target value (31/12/2012) for Ni according to EU directives = 20 ng/m3. The average values in 2011-2012 are below the target value of 20 ng/m3. The average Ni concentration decreases significantly from 2010 to 2012. The average values in Herstal (2012) are below the target value of 20 ng/m 3 and significant lower than the concentrations measured in Genk. Ni is mainly present in fine dust as an oxidic compound (48-83%) for both locations in Herstal and Genk. The sulfidic and metallic Ni species are less than 14 % present. The soluble Ni fraction is within the 8-38% concentration range. 92-62 % of the total Ni is present in the insoluble form.

4.4.3. Platinum Platinum can be used as target element for estimating the contribution of traffic to fine dust (source recognition).

The total amount of fine dust is not always related to the amount of Ni, Cr(VI) and Pt. Therefore it is of great importance to perform both mass analysis and speciation on fine dust.

186

4.5. References Determination of hexavalent chromium in ambient air: A story of method induced Cr(III) oxidation Kristof Tirez et al. Atmospheric Environment 45 (2011) 5332-5341 Speciation of Nickel in airborne particulate matter by means of sequential extraction in a micro flow system and determination by graphite furnace atomic absorption spectometry and inductively coupled plasma mass spectrometry Lars Füchtjohann et al. J. Environ. Monit., 2001, 3, 681-687 Platinum Group Elements: A Challenge for Environmental Analytics A. Dubiella-Jackowska et al. Polish J. of Environ. Stud. Vol. 16, No. 3 (2007), 329-345 Platinum group elements (Pt, Pd, Rh) in airborne particulate matter in rural vs. urban areas of Germany: Concentrations and spatial patterns of distribution. Fathi Zereini et al. Science of the Total Environment 416 (2012) 261-268

187

APPENDIX Action 5: Speciation – Source recognition – health effects Cr speciation Cr(VI) concentrations Genk (2011-2012) – Herstal (2012) Cr(VI) average duplo's (Genk 2011) 25

ng/m³

20 15 10 5 0

21-Dec

17-Dec

13-Dec

9-Dec

5-Dec

1-Dec

27-Nov

23-Nov

19-Nov

15-Nov

11-Nov

7-Nov

3-Nov

30-Oct

26-Oct

22-Oct

18-Oct

14-Oct

10-Oct

Mean Cr(VI) concentration: 3.10 ng/m³ Cr(VI) average duplo's (Genk 2012) 7 6

ng/m³

5 4 3 2 1 0 4-Jul

27-Jun

20-Jun

188

13-Jun

6-Jun

30-May

23-May

16-May

9-May

2-May

Mean Cr(VI) concentration: 0.95 ng/m³

Cr(VI) average duplo's (Herstal 2012) 0,3 0,25

ng/m³

0,2 0,15 0,1 0,05 0

24/apr

14/apr

4/apr

25/mrt

15/mrt

5/mrt

Mean Cr(VI) concentration: 0.11 ng/m³ Figures 1-2-3. Cr(VI) concentrations (ng/m³) measured in Genk (2011-2012) and Herstal (2012).

Cr(VI) average duplo's (Genk 2011) 25

ng/m³

20

15

10

5

0

21-Dec

17-Dec

13-Dec

9-Dec

5-Dec

1-Dec

27-Nov

23-Nov

19-Nov

15-Nov

11-Nov

7-Nov

3-Nov

30-Oct

26-Oct

22-Oct

18-Oct

14-Oct

10-Oct

189

Cr(VI) average duplo's (Genk 2012) 7 6

ng/m³

5 4 3 2 1 0

4-Jul

27-Jun

20-Jun

13-Jun

6-Jun

30-May

23-May

16-May

9-May

2-May

Cr(VI) average duplo's (Herstal 2012) 0,35 0,3 0,25

ng/m³

0,2 0,15 0,1 0,05 0

23-Apr

190

16-Apr

9-Apr

2-Apr

26-Mar

19-Mar

12-Mar

5-Mar

Figures 4-5-6. Average Cr(VI) concentrations (ng/m³) with the standard deviation.

Tables 2-3-4. Conversions between Cr species for Genk (2011-2012) and Herstal (2012).

GENK 2011

10-Oct 11-Oct 12-Oct 13-Oct 17-Oct 18-Oct 19-Oct 20-Oct 24-Oct 25-Oct 26-Oct 27-Oct 7-Nov 8-Nov 9-Nov 14-Nov 15-Nov 16-Nov 17-Nov 21-Nov 22-Nov 23-Nov 24-Nov 28-Nov 29-Nov 30-Nov 1-Dec 5-Dec 6-Dec 7-Dec 8-Dec 12-Dec 13-Dec 14-Dec 15-Dec 19-Dec 20-Dec 21-Dec

GENK 2012 2-May 3-May 7-May

average duplo1-2 8,85 6,55 1,98 0,08 3,25 9,35 4,37 0,97 0,07 0,21 0,23 0,13 < 0,05 < 0,05 0,26 0,11 0,04 0,97 0,17 0,27 0,43 0,16 0,29 0,09 1,94 1,17 4,71 19,18 16,49 11,14 20,16 0,73 0,27 0,85 0,44 2,30 0,27 0,03

theoretical average + spike Cr(VI) 9,64 6,93 2,36 0,46 3,63 9,73 4,75 1,54 0,45 0,59 0,61 0,52 0,65 0,71 0,82 0,49 0,83 1,87 0,96 1,27 1,89 0,97 1,08 0,88 2,73 1,96 5,50 19,98 17,28 11,94 20,95 1,52 1,06 1,65 1,23 3,09 1,06 0,82

average duplo1-2 0,84 0,77 0,12

theoretical average + measured % recovery spike Cr(VI) spike Cr(VI) Cr(VI) 1,05 0,98 0,91 0,68 71

measured spike Cr(VI) 8,33 6,38 2,28 0,33 3,45 9,24 4,78 1,34 0,33 0,39 0,49 0,28 0,42 0,42 0,69 0,36 0,63 1,74 0,70 0,88 1,55 0,61 0,63 0,38 2,23 1,03 4,51 16,02 15,29 10,95 20,09 1,40 0,92 1,66 1,12 2,75 0,84 0,61

191

% recovery Cr(VI) -65 -45 80 67 53 -29 108 64 69 48 69 40 66 60 77 67 75 86 66 61 77 55 43 36 37 -18 -24 -400 -151 -25 -9 85 81 102 86 57 72 73

measured spike Cr(III) 8,46 6,08 2,46 0,32 4,08 9,68 4,45 1,68

% Cr(III) oxidation -0,97 -2,48 2,51 1,27 4,36 1,74 0,42 2,34

0,63 0,31 0,45 0,48 0,30 0,42 0,59 0,64 1,80 0,52 1,44 1,64 0,49 0,77 0,37 1,86 1,02 4,68 18,15 17,04 11,50 22,25 1,14 0,57 1,34 0,87 2,77 0,80 0,48

1,98 0,40 1,70 1,52 0,81 0,54 2,52 1,54 1,88 0,90 1,82 1,49 0,66 1,22 0,70 -0,21 -0,37 -0,05 -2,61 1,40 0,90 5,28 1,03 0,75 1,22 1,10 1,19 1,34 1,12

measured spike Cr(III) 1,55 1,17 0,71

% Cr(III) oxidation 1,81 1,03 1,50

8-May 9-May 10-May 14-May 15-May 21-May 22-May 23-May 24-May 29-May 30-May 31-May 4-Jun 5-Jun 6-Jun 7-Jun 11-Jun 12-Jun 13-Jun 14-Jun 18-Jun 19-Jun 20-Jun 21-Jun 25-Jun 26-Jun 27-Jun 28-Jun 3-Jul 4-Jul 5-Jul

HERSTAL 2012 5-Mar 6-Mar 7-Mar 8-Mar 12-Mar 13-Mar 14-Mar 15-Mar 19-Mar 20-Mar

1,21 1,70 2,55 2,54 2,67 0,02 0,09 0,30 0,01 0,09 0,08 3,20 6,18 0,51 4,10 0,39 1,00 0,03 0,47 0,07 0,75 0,02 0,03 0,57 0,15 0,10 0,18 0,13 0,41 0,23 0,92

2,00 2,50 3,34 3,33 3,46 0,81 0,88 1,09 0,80 0,88 0,87 3,99 6,97 1,30 4,89 1,18 1,79 0,82 1,26 0,86 1,54 0,81 0,83 1,36 0,94 0,89 0,97 0,93 1,20 1,02 1,71

1,92 2,33 3,03 3,06 3,10 0,29 0,23 0,66 0,45 0,40 0,38 3,71 6,08 1,01 4,89 0,97 1,50 0,51 1,00 0,55 1,07 0,33 0,29 1,09 0,45 0,58 0,66 0,72 1,03 0,73 1,67

89 79 61 66 55 35 18 45 55 39 39 65 -12 64 99 73 64 61 67 62 40 40 32 66 38 60 60 74 79 63 94

1,68 2,23 3,33 3,01 3,15 0,34 0,41 0,51 0,24 0,26 1,18 4,17 7,61 0,88 5,27 0,78 1,62 0,37 0,70 0,46 0,98 0,29 0,54 1,02 0,33 0,51 0,40 0,78 0,95 0,84 1,64

1,17 1,34 1,96 1,19 1,22 0,80 0,81 0,54 0,58 0,44 2,79 2,47 3,62 0,94 2,94 0,98 1,57 0,87 0,58 0,99 0,58 0,69 1,28 1,14 0,45 1,04 0,55 1,63 1,36 1,56 1,83

average duplo1-2 0,03 0,15 0,05 0,10 0,14 0,11 0,16 0,17 0,09 0,08

theoretical average + spike Cr(VI) 0,82 0,94 0,84 0,89 0,93 0,90 0,95 0,96 0,88 0,87

measured spike Cr(VI) 1,20 0,57 0,59 0,64 0,46 0,47 0,36 0,50 0,62 0,60

% recovery Cr(VI) 148 53 68 68 41 45 25 41 68 65

measured spike Cr(III) 0,75 0,68 0,88 0,51 0,35 0,35 0,48 0,26 0,54 0,40

% Cr(III) oxidation 1,83 1,35 2,10 1,05 0,53 0,59 0,80 0,23 1,16 0,79

192

21-Mar 22-Mar 26-Mar 27-Mar 28-Mar 29-Mar 2-Apr 3-Apr 4-Apr 5-Apr 10-Apr 11-Apr 12-Apr 16-Apr 17-Apr 18-Apr 19-Apr 23-Apr 24-Apr 25-Apr 26-Apr

0,28 0,18 0,08 0,19 0,12 0,05 0,02 0,08 0,11 0,06 0,12 0,04 0,08 0,17 0,05 < 0.02 < 0.02 < 0.02 0,05 < 0.02 < 0.02

1,09 0,97 0,87 1,10 0,91 0,84 0,81 0,89 0,90 0,85 0,91 0,83 0,88 0,96 0,84 0,79 0,79 0,79 0,84 0,79 0,79

0,55 0,67 0,63 0,78 0,60 0,56 0,58 0,69 0,47 0,65 0,80 0,66 0,57 0,66 0,49 0,45 0,48 0,48 0,55 0,64 0,62

34 62 70 65 61 65 70 75 46 74 86 78 61 61 56 57 61 61 63 81 78

0,54 0,41 0,41 0,41 0,34 0,21 0,44 0,51 0,46 0,58 0,73 0,70 0,56 1,11 0,40 0,39 0,30 0,41 0,34 0,55 0,31

0,62 0,59 0,85 0,49 0,56 0,41 1,06 1,05 0,88 1,32 1,54 1,66 1,21 2,37 0,89 0,99 0,76 1,05 0,73 1,39 0,78

Genk 2011 23 20

Cr(VI) ng/m³

18 15 13 10 8 5 3 0 21-Dec

19-Dec

17-Dec

15-Dec

13-Dec

11-Dec

9-Dec

7-Dec

5-Dec

3-Dec

1-Dec

29-Nov

spike Cr(III)

27-Nov

25-Nov

23-Nov

193

21-Nov

spike Cr(VI)

19-Nov

17-Nov

15-Nov

13-Nov

11-Nov

9-Nov

7-Nov

5-Nov

3-Nov

1-Nov

30-Oct

28-Oct

26-Oct

24-Oct

22-Oct

20-Oct

18-Oct

16-Oct

14-Oct

12-Oct

10-Oct

Cr(VI)

1,2

0,8

0,6

0,4

0,2

0

5-Mar

7-Mar

9-Mar

11-Mar

13-Mar

15-Mar

17-Mar

19-Mar

21-Mar

23-Mar

25-Mar

27-Mar

29-Mar

31-Mar

2-Apr

4-Apr

6-Apr

8-Apr

10-Apr

12-Apr

14-Apr

16-Apr

18-Apr

20-Apr

22-Apr

26-Apr

1,4

5-Jul

Herstal 2012

24-Apr

1-Jul 29-Jun 27-Jun 25-Jun 23-Jun 21-Jun 19-Jun 17-Jun 15-Jun 13-Jun 11-Jun 9-Jun 7-Jun 5-Jun 3-Jun 1-Jun 30-May 28-May 26-May 24-May 22-May 20-May 18-May 16-May 14-May 12-May 10-May 8-May 6-May 4-May 2-May

194

spike Cr(III)

spike Cr(VI)

Cr(VI)

spike Cr(III) Cr(VI)

spike Cr(VI)

3-Jul

1

Cr(VI) ng/m³

6

5

4 Cr(VI) ng/m³

Genk 2012 8

7

3

2

1

0

Figures 7-8-9. The average Cr(VI) concentration and the concentration of Cr(VI) on the spiked filters with Cr(VI) and Cr(III).

Tables 5-6-7. Cr(VI)/Cr ratio (%) for Genk (2011-2012) and Herstal (2012).

Genk 2011 10-Oct 11-Oct 12-Oct 13-Oct 17-Oct 18-Oct 19-Oct 20-Oct 24-Oct 25-Oct 26-Oct 27-Oct 7-Nov 8-Nov 9-Nov 14-Nov 15-Nov 16-Nov 17-Nov 21-Nov 22-Nov 23-Nov 24-Nov 28-Nov 29-Nov 30-Nov 1-Dec 5-Dec 6-Dec 7-Dec 8-Dec 12-Dec 13-Dec 14-Dec 15-Dec 19-Dec 20-Dec 21-Dec

Genk 2012 2-May 3-May 7-May

Average ng Cr(VI)/m 8,85 6,55 1,98 0,08 3,25 9,35 4,37 0,97 0,07 0,21 0,23 0,13 < 0,05 < 0,05 0,26 0,11 0,04 0,97 0,17 0,27 0,43 0,16 0,29 0,09 1,94 1,17 4,71 19,18 16,49 11,14 20,16 0,73 0,27 0,85 0,44 2,30 0,27 0,03

3

Average ng Cr(VI)/m 0,84 0,77 0,12

3

Total ng Cr/m3 (VMM) 319 483 197 14,2 54,6 204 215 85,1 2,9 4,2 21,9 5 5,4 6,5 13 17 7,8 17,3 30,7 8,5 23,5 31,6 15,4 5,5 20,4 58,1 45,8 206 265 187 255 86,6 20,4 51,8 43,5 11,4 67,8 4

Cr(VI)/Cr ratio in %

Total ng Cr/m3 (VMM) 13,6 69,9 6,4

Cr(VI)/Cr ratio in %

195

2,77 1,36 1,01 0,55 5,95 4,58 2,03 1,14 2,45 5,04 1,06 2,54