Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon,...

45
Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris, France
  • date post

    19-Dec-2015
  • Category

    Documents

  • view

    213
  • download

    0

Transcript of Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon,...

Page 1: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Particulate Matter Modeling:Scientific Issues & Future Prospects

Christian Seigneur

AER

San Ramon, California, USA

EMEP TFMM, 29 November 2006, Paris, France

Page 2: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Major componentsof a PM air quality model

Initial and Boundary Conditions

MeteorologicalModel Emissions

Concentrations of gases and PM

DropletChemistry

WetDeposition

Dry Deposition

Gas-phasechemistry

PM Chemistry and Physics

TransportAir Quality PM Model

Page 3: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Model uncertainties:Inputs and formulation

• Inputs– Emissions – Meteorology– Boundary and initial conditions

• Formulation (some current issues)– Dry deposition velocities– Treatment of point sources– Treatment of secondary organic aerosols (SOA)

Page 4: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Emissions of PM and precursors

Bottom-up emission inventories are improving with inventories at highspatial and temporal resolution becoming more common:

– United Kingdom: National atmospheric emissions inventory (www.naei.org.uk)

– France: “Inventaire national spatialisé” (INS) under development by the French Ministry of Ecology (to be available in 2008)

PM models require accurate emissions of gaseous precursors (NOx, VOC, SO2, NH3) and primary PM (with size distribution and chemical composition)

Page 5: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Importance of NH3 emissions

Particulate nitrate concentrations decrease significantly when NH3 emissions decrease by 33%

Simulation of the eastern U.S. on 15 January 2002 with CMAQ

Page 6: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Meteorology

• Winds affect the accuracy of long-range transport

• Vertical mixing affects PM surface concentrations

• Clouds (and fogs) enhance secondary PM formation but precipitation removes PM from the atmosphere

Page 7: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Model performance for transport

Regional models cannot reliably predict the impact of individual point sources at long distances

Comparison of measured and simulated concentrations of a tracer released at 500 km from the receptor during July-October 1999 in Texas (Pun et al., JGR,, 111, D06302, doi:10.1029/2004JD005608, 2006)

Page 8: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Comparison of two PM modelsImportance of vertical mixing

PM2.5 concentrations over the U.S. on 6 July 1999 differ

primarily because of different algorithms for vertical mixing

(Zhang et al., JAWMA, 54, 1478, 2004)

CMAQ CAMx

Page 9: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Influence of clouds on sulfate: Formation and removal

Without clouds With clouds

Simulation of 14 July 1995 with CMAQ over the northeastern U.S.

Page 10: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Importance of boundary conditions

Sulfate over the United States

Simulation of July-October 1999 (REMSAD, M. Barna, National Park Service)

Solution: Use of a global model that has undergone satisfactory performance evaluation

Page 11: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Importance of the dry deposition of HNO3

Particulate nitrate concentrations decrease significantly when the dry deposition velocity of HNO3

increases by a factor of 3

Simulation of the eastern U.S. on 15 January 2002 with CMAQ

Page 12: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Evolution of plume chemistry

Early Plume Dispersion

NO/NO2/O3 chemistry

1

2

Mid-range Plume Dispersion

Reduced VOC/NOx/O3 chemistry — acid formation from OH and NO3/N2O5 chemistry

Long-range Plume Dispersion

3

Full VOC/NOx/O3 chemistry — acid and O3 formation

PM formation is negligible near the stack

(Karamchandani et al., ES&T, 32, 1709,1998)

Page 13: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Effect of an advanced plume-in-grid treatment (APT) on sulfate concentrations

Without APT With APT

Simulation of July 2002 over the eastern United States with CMAQ-MADRID: Sulfate due to 14 power plants

(Karamchandani et al., AE, 40, 7280, 2006)

Page 14: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Chemical composition of PM:Organics constitute a significant fraction

PM10, Saclay, Paris region, France, July 2000

(Hodzic et al., ACP, 6, 3257, 2006)

Other

Black Carbon

Organic Mass

Ammonium

Nitrate

Sulfate

PM2.5, Boundary Waters, Minnesota, USA, July 2002

(IMPROVE network)

Page 15: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Treatment of SOA formation

Cgas

Cparticle

K

Cgas

Cparticle

H

Condensable VOC oxidation products

Hydrophilic Hydrophobic

Henry’s law for the aqueous phase

Raoult’s law for the organic phase

Page 16: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Uncertainties in SOA formation

• Missing precursors– Isoprene, benzene, sesquiterpenes are now being added

to models• Large number of condensable products

– Use of surrogate SOA compounds• Approximations for the partitioning constants• Oligomerization

– Not currently treated in models• Limited interactions between organic and inorganic

compounds in particles

Page 17: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Effect of oligomerization on organic PM

Effect of pH on SOA oligomerization in an -pinene simulation(Pun and Seigneur, www.crcao.com, 2006)

0

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

3 4 5 6 7

pH

SO

A Y

ield

(

M/

HC

)

0

2

4

6

8

10

12

14

16

C(

g/m

-3 )

Y

SOA mass

Oligomer mass

Page 18: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Model performance evaluation

Model application

Model evaluation

Model/Data improvement

Regulatoryapplication

or forecasting

The modeling cycle iterates until performance is good enough for emission strategy design or forecasting

Page 19: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Model performance evaluation: Comparing the model results to data

Ambient Data Model Results

Model Evaluation Software + Graphics Package

Performance StatisticsPaired peak errorUnpaired peak errorGross errorGross biasNormalized biasNormalized errorRoot mean square errorCoefficient of determination...

GraphicsTime series

Scatter plots

Pie charts

...

Observed PM2.5 =6.49 µg/m3

EC1.8%

OC19.5%

Other24.1%

SO42–

38.9%

NO3–

3.1%NH4

+

12.6%

EC1.8%

OC19.5%

Other24.1%

SO42–

38.9%

NO3–

3.1%NH4

+

12.6%

Predicted PM2.5 =4.66 µg/m3

EC1.3%

Other17.2%

OC9.5%

NH4+

14.7%

SO42–

56.4%

NO3–

0.8%

EC1.3%

Other17.2%

OC9.5%

NH4+

14.7%

SO42–

56.4%

NO3–

0.8%

0

1

2

3

4

5

6

7/2 7/9 7/16 7/23 7/30 8/6 8/13 8/20 8/27 9/3 9/10 9/17 9/24 10/1 10/8 10/15 10/22

Total Sulfur (

/g m

3)

Big Bend

y = 0.22x + 2.95

R2 = 0.09

0

5

10

15

20

0 5 10 15 20

Observed Sulfate ( /g m3)

Simulated Values

Page 20: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Major issues when comparing models and measurements

• Spatial averaging• Temporal averaging• PM size fractions• PM chemical composition

– Semi-volatile species– Carbonaceous species– “Other” PM

Page 21: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Spatial averaging

• Spatial variability for a primary pollutant can be up to a factor of 2.5 (maximum/minimum) for a grid resolution of 4 km

• It will be less for a secondary pollutant

Point measurement

Model grid average

+

Page 22: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Temporal averaging

• Models and measurements are consistent for short periods (1 to 24-hour averaging)

• Lack of daily measurements may lead to approximations of seasonal and annual measured values (e.g., 1 in 3 days for U.S. networks)

• It is preferable to conduct model performance evaluations using continuous measurements with fine temporal resolution (~ 1 hour)

Page 23: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Diagnostic analysis using fine temporal resolution

Observed and simulated (CMAQ-MADRID 2) organic mass

in Atlanta, Georgia, USA, July 1999

(Bailey et al., JGR, in press)

0

5

10

15

20

6/30/99 0:00 7/2/99 0:00 7/4/99 0:00 7/6/99 0:00 7/8/99 0:00 7/10/99 0:00

Local Time

Organic Mass (

/g m

3)

1.4OCx 8km

Page 24: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Sampling PM2.5

Measurements do not have a sharp particle diameter cut-off: PM2.5 includes some coarse particles and some fine particles are not sampled.

Page 25: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

PM size fraction

• Inertial impaction measurements use the aerodynamic diameter of the particles to define the size fraction

– the aerodynamic diameter, da, is the diameter of a spherical particle of unit density that behaves like the actual particle

• Models simulate particle dynamics using the Stokes diameter

– the Stokes diameter, dS, is the diameter of a spherical particle that behaves like the actual particle

Page 26: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

PM diameters

dS = da / (particle density)1/2

Particle density is a function of location and time

For example, if one uses an average PM2.5 density of 1.35 g/cm3, dS of PM2.5 in the model is 2.15 m

Page 27: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

PM2.5 chemical composition

Nitrate: sampling artifacts

Sulfate

Ammonium: sampling artifacts

Organics: Conversion factor OC => OM? Sampling artifacts

BC: factor of 2?

Other: some volatilization? some water?

Page 28: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Semi-volatile species

HNO3 & nitrate

NH3 & ammonium

Organic compounds Water

Their particulate mass can be over- or underestimated due to positive or negative artifacts

Page 29: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Ammonium nitrate

• Positive artifacts may occur in the absence of upstream denuders to collect gaseous HNO3 and NH3

• Negative artifacts may occur because of evaporation of ammonium and nitrate from the filter – Significant for Teflon filters (losses are typically

higher in summer)– Nitrate is thought to be well collected on Nylon filters

impregnated with alkaline substance but some ammonium could be volatilized

Page 30: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Organic compounds

• Positive artifacts may occur because denuders are rarely used (adsorption of gaseous organic compounds on quartz filters)

• Negative artifacts may occur due to volatilization of organic PM (it can be significant: about 50% in Riverside, California, according to Pang et al., AST, 36, 277, 2002)

Page 31: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Carbonaceous species

The difference between black carbon (BC) and organic carbon (OC) is operational:

- temperature evolution protocol

- light absorption method

Different monitoring networks use different techniques

~factor of 2 difference for BC (Chow et al., AST, 35, 23, 2001)

~10% difference for OC

For modeling, the emissions and ambient determinations of BC should be based on the same operational technique

Page 32: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Estimating organic PM

Organic mass is not measured but estimated from measured organic carbon using a scaling factor

– the default value is 1.4

– it can range from 1.2 to 2.6

Turpin and Lim (AST, 35, 602, 2001) recommend

– 1.6 for urban PM

– 2.1 for non-urban PM

Page 33: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Advanced techniques for evaluating model performance

• Evaluating spatial and temporal patterns• Evaluating the third dimension• Comparing different models and/or modeling

techniques• Evaluating model response

Page 34: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Spatial display of model error can provide insights into possible causes

< 1515 to 3030 to 4545 to 6060 to 7575 to 9090 to 105105 to 120> 120

%58.6

76.6

54.5

69.7

63

69.4

97.4

52.1

80.6

70

65.7 58.854.6

32.6

40.5

78.5

49

158.7

54

78.6

63.240.351.1

36.4

140

40.8

72.7

55.9

48.6

7749.8

90.369.2

49.7

67

36.7

66.2

CMAQ Sulfate StatisticsNormalized Error (%)

< 1515 to 3030 to 4545 to 6060 to 7575 to 9090 to 105105 to 120

< 1515 to 3030 to 4545 to 6060 to 7575 to 9090 to 105105 to 120> 120

%58.6

76.6

54.5

69.7

63

69.4

97.4

52.1

80.6

70

65.7 58.854.6

32.6

40.5

78.5

49

158.7

54

78.6

63.240.351.1

36.4

140

40.8

72.7

55.9

48.6

7749.8

90.369.2

49.7

67

36.7

66.2

CMAQ Sulfate StatisticsNormalized Error (%)

Sulfate error for

CMAQ-MADRID

in Texas

(July-October 1999):

- Emissions ?

- Coastal meteorology ?

(Pun et al., JGR,, 111, D06302, doi:10.1029/2004JD005608, 2006)

Page 35: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Evaluating the third dimensionUse of satellite data: Aerosol Optical Depth

August 2001 (Bias = -51%) November 2001 (Bias = -35%)

Model simulation:CMAQ

Satellite data:MODIS

(Zhang et al., AMS annual meeting, 2006)

Page 36: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Comparison of three SOA models

Odum/Griffin

0

0.05

0.1

0.15

0.2

0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68Time (hr)

SOA (

/g m

3 )BiogenicAnthropogenic

/CMU STI

0

0.5

1

1.5

2

0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68

( )Time hr

(SOA

/g m

3 )BiogenicAnthropogenic

AEC

0

5

10

15

0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 ( )Time hr

(SOA

/g m

3 )BiogenicAnthropogenic

 The three SOA models differ in:

• the total amounts of SVOC and SOA

• the gas/particle partitioning

• the relative amounts of anthropogenic and biogenic SOA

(Pun et al., ES&T, 37, 3647, 2003)

Page 37: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Reconciliation of CTM and receptor modeling results

Refinement of the modeling results using additional information leads to better agreement among the different models (Pitchford et al., BRAVO report, vista.cira.colostate.edu/improve, 2004)

Contribution of various source areas (Mexico, Texas, eastern U.S. and western U.S.) to sulfate in Big Bend National Park, Texas

Page 38: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Response of PM to changes in precursors

Change in PM compositionReduction inemissions Sulfate Nitrate Organics

SO2

NOx

VOC

NH3

Black carbon

Primary OC

(adapted from Pandis, www.narsto.org, 2004)

Page 39: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Effect of decreased precursor emissionson PM concentrations (g/m3)

Sulfate

Nitrate

Organics

SO2 NOx VOC

(Seigneur, AIChEJ, 51, 355, 2005)

Page 40: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Evaluating model response

• A satisfactory operational evaluation does not imply that a model will predict the correct response to changes in precursors emissions

• There is a need to conduct a diagnostic/mechanistic evaluation to ensure that the model predicts the correct chemical regimes

• Indicator species can be used to evaluate the model’s ability to predict chemical regimes

Page 41: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Major chemical regimes

• Sulfate

– SO2 vs. oxidant-limited

• Ammonium nitrate

– NH3 vs. HNO3-limited

• Organics

– Primary vs. secondary

– Biogenic vs. anthropogenic

• Oxidants (O3 & H2O2)

– NOx vs. VOC-limited

Page 42: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Example of indicator speciesSensitivity of O3 formation to VOC & NOx

• H2O2 / (HNO3 + Nitrate) as an indicator

Low values: VOC sensitive

High values:NOx sensitive O3

NO NO2

HNO3

OHHO2H2O2

VOC

Page 43: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Example of indicator speciesSensitivity of nitrate formation to

NH3 & HNO3

• Excess NH3 as an indicator

Low values: NH3 sensitive

High values:HNO3 sensitive

Ammonium nitrate

HNO3 NH3

Ammonium sulfate

Page 44: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Qualitative estimates of uncertainties

Component Confidencea Major uncertainties

Sulfate M-H Clouds & precipitation

Nitrate L-M Emissions & partitioning

Ammonium L-M Emissions

Primary OC L Emissions

Secondary OC VL VOC emissions & formation

BC L Emissions

Crustal & seasalt L Emissions

Others VL Emissions

(a) H: high, M: medium, L: low, VL: very low (adapted from Seigneur & Moran, www.narsto.org, 2004)

Page 45: Particulate Matter Modeling: Scientific Issues & Future Prospects Christian Seigneur AER San Ramon, California, USA EMEP TFMM, 29 November 2006, Paris,

Possible topics for improvingPM model performance

• Model inputs– Emission inventories (ammonia, primary PM, etc.)– Transport processes (e.g., vertical mixing)– Assimilation of cloud and precipitation data– Boundary conditions from global models

• Model formulation– Advanced plume-in-grid treatment for point sources– SOA formation– Heterogeneous chemistry– Deposition velocities