Current practice of PM-measurements, data processing, interpretation and visualization in Belgium Frans Fierens scientific staff member of the Flemish.

Current practice of PM-measurements, data processing, interpretation and

visualization in Belgium

Frans Fierensscientific staff member of the Flemish Environment Agency

(VMM) at the Belgian Interregional Environment Agency (IRCEL)

PM_lab workshop, 2010 March 4

IRCEL-CELINE ?

NL : Intergewestelijke Cel voor het LeefmilieuFR : Cellule Interrégionale de l'Environnement

EN : Belgian Interregional Environment Agency

Agreement between the 3 Belgian Regions (1994)

• Major tasks :• SMOG (winter/summer) warnings (IDPC)• Interregional Calibration Bench• Interregional AQ Database (3 Regions)• Scientific support• Reports EU-COM / Experts EU-working groups

Contents

1. Choice of PM-Measurement locations

2. Calibration of PM-Measurements - equipment

3. Future technical development in the next 2-3 years

4. Data acquisition - Handling of PM-data

5. Spatial Interpolation of PM-point data

6. Forecast Modelling (deterministic / statistical models).

Contents







Number of PM10 and PM2.5 monitoring stations

• PM10 : start measurements in 1996• PM2.5 : start measurements in 2000

PM10 (telemetric stations)(>90% valid daily averages)

0

10

20

30

40

50

60

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

# P

M10

sta

tio

ns

Wallonia

Brussels

Flanders

0

10

20

30

40

50

60

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009#

PM

2.5

stat

ion

s

PM2.5 (telemetric stations)(>90% valid daily averages)

Beside PM : also BC and Black Smoke measurements

PM10 : monitoring stations

Location of PM10 telemetric stations

Locations mostly :- Industrial- Urban or Urban Background

Very few “rural” and “traffic” stations

(Historical reasons)

PM2.5 : monitoring stations

Location of PM2.5 telemetric stations

Locations mostly :- Industrial- (Sub) Urban

Very few “rural” and “traffic”

“AEI stations” :-Bruges(*)-Ghent (*)-Antwerp : 2 (*)-Brussels : 2-Liège-Charleroi

(*) not on the map

Contents







PM measuring techniques in Belgium

1. Flanders- Oscillating Micro Balans (TEOM and TEOM-FDMS)- Bèta Absorption (ESM FH62I-R)- Gravimetric :

- Equivalence tests- PM2.5 (to calculate the Average Exposure Index “AEI” on urban-

background locations, started in 2009) + 1 Rural background location

2. Brussels- Oscillating Micro Balans (only TEOM-FDMS since 2004-2005)

3. Wallonia- Bèta Absorption (MP101 integration time 24h)- Optical techniques (GRIMM)

Automatic PM monitors <> EU reference method

PROBLEM : automatic monitors <> EU (gravimetric) reference method

NO PROBLEM :When “equivalence” is demonstrated

Current “calibration” of PM in Belgium

(*) based on the ‘guide for the demonstration of equivalence of ambient air monitoring methods’ (Excel templates from the JRC)

(**) preliminary results of an equivalence program in Wallonia result in somewhat higher calibration factors

PM10 currently used calibration factors Equivalence after "correction" (*)ESM 1.37 yesTEOM 1.47 yesTEOM-FDMS 1 yesGRIMM 1 yes (**)

PM2.5 currently used calibration factors Equivalence after "correction" (*)ESM 1.46 yes TEOM 1.75 yes TEOM-FDMS 1 yes GRIMM 0.85 yes (**)

New comparative campaign (VMM) : PM10

“calibration” factors calculated in new campaign are slightly higher than previously

“Comparative PM10 and PM2.5 measurements in Flanders (Belgium)”, VMM, Period 2006 - 2007 (www.vmm.be)

First comparative campaign (VMM) : PM2.5

Higher “calibration” factors for PM2.5 than for PM10

-> higher volatile fraction

“Comparative PM10 and PM2.5 measurements in Flanders (Belgium)”, VMM, Period 2006 - 2007 (www.vmm.be)

Spatial and temporal variation of calibration factors

Contents







Future technical development in the next 2-3 years (1)

Flanders :- More “Chemkar” campaigns ( PM10 “hotspots”,Rural vs Urban PM10 & PM2.5, Antwerp harbour, …)- Measuring the effect of Woodburning on PM (levoglucosan)- Additional measuring stations (e.g. Streetcanyon NO2/PM)- Testing of new Bèta-monitors (BAM1020, FAI SWAM 5DC)- UFP measurements (streets) - Further participating in CEN/TC264/WG15 :

* revision of the PM10 standard EN12341* revision of the PM2.5 standard EN14907

Future technical development in the next 2-3 years (2)

Brussels :- “Black Carbon” measurements- “Counting Particles” (using GRIMM monitors)

Wallonia :- additional measuring stations (e.g. Tournai, Namur)- EC/OC analyser at Vielsalm (Rural background)

Interregional (IRCEL-CELINE) :- further developing Interpolation techniques (eg. use of satellite observations like AOD)- higher spatial resolution modelling (forecasts + assessment)- implementation of data assessment techniques

Contents







Data acquisition of automatic measurements

Monitoringstation

RDRC

IRCEL

Every hour (26’ after each hour)

-> ½ - hourly measurements

-> FTP to IRCEL servers

-> calculation of hourly / 8-hourly / 24-hour averages.

-> publication real-time data + maps on websites

“Regional Data Processing Centers”

“Real-Time” publication on websites - tables

“Real-Time” publication on websites - maps

Contents







How to define a scientifically based methodology for assessment of spatial representativeness?

CORINE land use map

• Observation:• Sampling values depend on land use in (direct)

vicinity of the monitoring site

• Consequence:• Interpolation scheme needs to know this relation

between land use and air quality levels

• Approach :• Create land use indicator to express this relation

RIO-Corine interpolation

VITO + IRCEL developed theRIO-corine methodology

43R240

2 km

0 5 10 15 20 25 30 35 400

50

100

150

200

250

300

350

400

CORINE Class

Num

ber

of g

rid c

ells

in b

uffer

42N016

43N073

43R240

Land use indicatorFor each station: Determine buffer

(e.g. 2km radius) Characterize land

use by CORINE class distribution inside buffer

RIO - Land use indicator (1)

Land use indicator is based on CORINE class

distribution

Calibration of coefficients ai :

multi-regression to optimize trend for mean and standard dev. of monitoring data

RIO - Land use indicator (2)

0 0.5 1 1.50

20

40

60

80

100

120

140

NO

2 [

g/m

3 ]

week

rural

urb back

urbind

traff

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.110

15

20

25

30

35

40

45

50

55

60

PM

10 [

g/m

3 ]

week

rural

urb back

urbind

traff

<PM10>

<NO2>

‘Kriging’ condition = ‘spatialy’ homogeneous data

Use relation between land use indicator and AQ statistics to “detrend” monitoring data:

Remove local character of sampling values

Kriging interpolation of “detrended” data

0 0.2 0.4 0.6 0.8 1 1.2 1.40

10

20

30

40

50

60

70

80

90

C

[µg/

m³]

C

1. Detrend sampling values

2. Interpolate detrended values with Ordinary Kriging

3. Determine local -value

4. Get corresponding trend shift (C)

5. Add C to interpolation result

RIO-corine methodology

LegendRIO map [µg/m³]

34 - 43

44 - 52

53 - 61

62 - 68

69 - 75

76 - 126

LegendKriging [µg/m³]

78 - 85

86 - 87

88 - 90

91 - 95

96 - 99

100 - 109

Legenddelta C [µg/m³]

-4 - 0

1 - 21

22 - 27

28 - 33

34 - 39

40 - 52

(1) Kriging map (2) DC map

(3) RIO map

0 0.2 0.4 0.6 0.8 1 1.2 1.40

10

20

30

40

50

60

70

80

90

C

[µg/

m³] C

Correlation <-> distance

Compare with standard IDW and OK

Valdidation – “leaving-one-out”

Model O3 NO2 PM10

RMSE

Bias RMSE Bias RMSE Bias

IDW 10.97 -1.70 18.17 4.74 12.12 1.70

OK 10.37 -0.44 16.85 1.45 11.65 1.22

RIO 9.56 -0.08 14.45 -0.67 9.89 0.01

Valdidation – using “independent” measurements

0

10

20

30

40

50

60

70

80

90

20

04

06

26

20

04

07

01

20

04

07

06

20

04

07

11

20

04

07

16

20

04

07

21

20

04

07

26

20

04

11

26

20

04

12

01

20

04

12

06

20

04

12

11

20

04

12

16

20

04

12

21

20

04

12

26

20

04

12

31

20

05

01

05

20

05

01

10

observations

RIO-corine interpolated

R² = 0.90MAE = 2.9 µg/m³RMS = 4.3 µg/m³

Average observations : 30.6 µg/m³Average RIO-c interpolation : 31.5 µg/m³

Annual mean PM10 concentrations 2006

RIO-corineOrdinary Kriging

RIOOK

Annual average NO2 concentrations 2002

LegendNO2 [µg/m³]

error

1 - 10

11 - 12

13 - 14

15 - 17

18 - 20

21 - 23

24 - 26

27 - 29

30 - 33

> 33

LegendNO2 [µg/m³]

error

1 - 10

11 - 12

13 - 14

15 - 17

18 - 20

21 - 23

24 - 26

27 - 29

30 - 33

> 33

LegendNO2 [µg/m³]

error

< 10

11 - 12

13 - 14

15 - 17

18 - 20

21 - 23

24 - 26

27 - 29

30 - 33

> 33

RIO-corine : further developments (1)

NO2 - 4x4 km NO2 - 1x1 km

RIO-corine : further developments (2)

New proxy :AOD (aerosol optical Depth) ?

Source : Modis Terra satelite, 2006

Total ColumnAOD 2006

RIO-corine : more info

“Spatial interpolation of air pollution measurements using CORINE landcover data ”Janssen Stijna, Dumont Gerwinb, Fierens Fransb, Mensink Clemensa

aFlemish Institute for Technological Research (VITO),Boeretang 200, B-2400 Mol, BelgiumbBelgian Interregional Cell Environment Agency(IRCEL), Kunstlaan 10-11, B-1210 Brussels, Belgium

Atmospheric Environment 42/20 (2008) 4884-4903

Contents







Brussels

Polluants : PM10 et NO2

Plan : 3 niveaux

Wallonia

Polluant : PM10

Plan : 3 niveaux

Flanders

Polluant : PM10

Plan : 1 niveau

- Information of the public (see ozone EU info/alert thresholds)

- Activation winter SMOG action plans

(FORECASTED PM10 > 70 µg/m³, for two consecutive days)

Goal of Air Quality forecasts ?

Two different types of models

1. Deterministic models• Complex input :

meteo, emissions, geografical information, fysico-chemical processes

• Long CPU

-> CHIMERE (forecasts) / BelEUROS (emission scenario’s)

2. Statistical or neural-network models• Simple input :

database with measurements, some simple forecasted meteo parameters

• Short CPU (minutes)

-> SMOGSTOP (Ozone) / OVL (PM10, NO2)

CHIMERE : simple schematic overview

NOx emissionscombustion

ExampleTemperature

CHIMERE – Example (1)

Forecast for 21/6/2005

Observations 21/6/2005

CHIMERE – Example (2)

OVL : schematically

Input:•PM10 measurements day-1

•Meteo forecasts

Process:

Output : PM10 daily mean day0, +1, +2, +3 and

+4

Neural

Network

OVL : most important meteo-input parameter

Temperature Inversion

Low windspeeds

Boundary Layer Height

OVL : PM10 – winter/spring 2005forecast day +1

0

20

40

60

80

100

120

140

01/01/05 01/02/05 01/03/05 01/04/05

µg/m

³

metingen

OVL modelR=0.7

Antwerp (monitoring station 42R801)

OVL : more info

“A neural network forecast for daily average PM10 concentrations in Belgium”Hooyberghs Jefa, Mensink Clemensa, Dumont Gerwinb, Fierens Fransb, Brasseur Olivierc

aFlemish Institute for Technological Research (VITO),Boeretang 200, B-2400 Mol, BelgiumbInterregional Cell for the Environment (IRCEL), Kunstlaan 10-11, B-1210 Brussels, BelgiumcRoyal Meteorological Institute (RMI), Ringlaan 3, B-1180 Brussels, Belgium

Atmospheric Environment 39/18 (2005) 3279-3289

Dank voor uw aandacht !

Je vous remercie de votre attention !

Wir danken Ihnen für Ihre Aufmerksamkeit !

Thank you for your attention !

More info :www.ircel.bewww.vmm.be

Current practice of PM-measurements, data processing, interpretation and visualization in Belgium Frans Fierens scientific staff member of the Flemish.

Documents

groupscontentschoice

mapcontentschoice of

current calibration

pm measuring techniques

current practice of

valid daily averagesbeside

data processing

urbanbackground locations