A Critical Evaluation of U.S. On-road Vehicle Emissions in ... 06... · David Parrish - NOAA ESRL ... VOC Speciation in U.S. on-road vehicle emissions: Characterization through ambient

David Parrish - NOAA ESRL

• Investigate CO/NOx ratios and CO levels in U.S. ambient data

Today:

A Critical Evaluation of U.S. On-road Vehicle Emissions in National and International Inventories

• Suggest needed improvements to U.S. on-road vehicle inventories and compare to international inventories

• Investigate possible application to Beijing

• Investigate VOC speciation through ambient data in U.S. and Beijing

Implicit message:We often hear that U.S. emission inventories are in pretty good

shape; that is not true in all respects (i.e. on-road emissions from MOBILE6).

Do not emulate U.S. techniques without critical evaluation.


Proposed Conclusion and Recommendation:It is not possible to accurately estimate emissions without input

from direct measurements (e.g. CEMS on point sources, satellite measurements, top-down tests through ambient measurements, inverse modeling, etc.)

An iterative process is required between inventory developers and measurers to reach accurate inventories.

1200

1000

800

600

400

200

0

CO

(ppb

v)

12:00 AM7/7/99

6:00 PM

150

100

50

0

NO

y (ppbv)

6:00 - 8:30

(5 min averages)

Ambient CO and NOycorrelate well duringmorning rush hour.

Measurementscharacterize CO/NOxemission ratio from on-road vehicles.

CO to NOx ratio in U.S. on-road vehicle emissions: Characterization through ambient measurements

Nashville,Tennessee

CO/NOx emission ratio from on-road vehicles

2500

2000

1500

1000

500

0

CO

(ppb

v)

200150100500NOy (ppbv)

Year Slope r2

1994 10.2 ± 1.5 0.97 1995 8.5 ± 1.3 0.90 1999 6.3 ± 0.9 0.95

6:00 - 8:30Nashville, TN(5 min averages)

• Ratio of CO to NOx from on-road vehicles in U.S. has decreased dramatically

3000

2000

1000

0

CO

(ppb

v)

16012080400NOy (ppbv)

Year Slope r2

1989 18.9 ± 2.7 0.95 1991 15.7 ± 2.3 0.94 1996 12.7 ± 2.0 0.87 1998 8.9 ± 1.3 0.94

6:00 - 10:00Boulder, CO

(5 min averages)

Temporal trends in CO to NOx ratios in U.S. urban areas

Morning rush hour dataUrban sites

7.9 ± 0.1

6.5 ± 0.4

5.7 ± 0.4

9.0 ± 1.2

9.4 ± 0.7

Ratio in 2000

0.97-6.6 ± 0.3AIRS

--------Atlanta

0.96-8.8 ± 1.0Nashville

0.94-6.7 ± 0.5Boulder

0.95-5.5 ± 0.4Los Angeles

r2Trend %/yrLocation

VOC/NOx emission ratiohas decreased similarly

Ratio of CO to NOx from on-road vehicles in U.S. has decreased dramatically

Parrish, “Critical evaluation of US on-road vehicle emission inventories” Atmospheric Environment, 40, 2288-2300, 2006.


MOBILE6 emission ratios have not decreased fast enough, and are too high now.

Temporal trends in CO to NOx ratios in U.S. urban areas

7.9 ± 0.1

6.5 ± 0.4

5.7 ± 0.4

9.0 ± 1.2

9.4 ± 0.7

Ratio in 2000

0.97-6.6 ± 0.3AIRS

--------Atlanta

0.96-8.8 ± 1.0Nashville

0.94-6.7 ± 0.5Boulder

0.95-5.5 ± 0.4Los Angeles



MOBILE6: Current EPA model

Recent data from Texas

Expanded axes, AIRS data as reference




CO

to N

Ox M

olar

Rat

io

2008200420001996

AIRS

El Paso

San Antonio

Dallas

Houston

MoodyTower

NEI1999

8

6

4

10

14


7.9 ± 0.1

7.3 ± 0.6

7.8 ± 0.7

9.1 ± 0.5

12.4 ± 0.7

Ratio in 2000

0.97-6.6 ± 0.3AIRS

0.86-5.8 ± 2.2Dallas

0.76-4.9 ± 2.5Houston

0.94-5.8 ± 1.5San Antonio

0.94-6.9 ± 1.7El Paso


Decrease continues through the present

Recent data from Texas



What drives the CO/NOx decrease?

Ambient urban CO levels have decreased: 4.6%/yr, but less rapidly than CO/NOxemission ratio: 6.6%/yr

NOx emissions must have increased!

Parrish, “Critical evaluation of US on-road vehicle emission inventories” AtmosphericEnvironment, 40, 2288-2300, 2006.

What drives the CO/NOx decrease?

Ambient urban CO levels have decreased: 4.6%/yr, but less rapidly than CO/NOxemission ratio: 6.6%/yr

NOx emissions must have increased! Not captured by MOBILE6

Parrish, “Critical evaluation of US on-road vehicle emission inventories” AtmosphericEnvironment, 40, 2288-2300, 2006.

MOBILE6 CO emissions about factor of 2 high, although rate of decrease well fit.

MOBILE6 VOC emissions have rate of decrease similar to CO, and are much more accurate

10

8

6

4

2

0

NO

x em

issi

ons

MOBILE5 NEI MOBILE6 NEI Inferred Emissions

UNECE/EMEP SNAP UNECE/EMEP NFR02

120

100

80

60

40

20

0

CO

em

issi

ons

200520001995199019851980

(units: Tg/year)

International View of U.S. On-Road Emissions

Comparison of MOBILE6 and Inferred Emissions with UNECE/EMEP and EDGAR Emission inventories

The U.S. officially submits data to the UNECE/EMEP database under the 1979 LRTAP Convention.

The current UNECE/EMEP data are not internally consistent, and subjectto the uncertainties discussed here.

10

8

6

4

2

0

NO

x em

issi

ons

MOBILE5 NEI MOBILE6 NEI Inferred Emissions

UNECE/EMEP SNAPUNECE/EMEP NFR02

EDGAR

120

100

80

60

40

20

0

CO

em

issi

ons

200520001995199019851980

(units: Tg/year)The EDGAR (Emission Database for

Global Atmospheric Research)inventory is a reference data base for many applications.

International View of U.S. On-Road Emissions

Comparison of MOBILE6 and Inferred Emissions with UNECE/EMEP and EDGAR Emission inventories

The current UNECE/EMEP data are not internally consistent, and subjectto the uncertainties discussed here.

The EDGAR inventory:• Underestimates NOx• Does not increase NOx fast enough• Overestimates CO• Does not decrease NOx fast enough

1200

1000

800

600

400

200

0

CO

(ppb

v)

12:00 AM7/7/99

6:00 PM

150

100

50

0N

Oy (ppbv)

6:00 - 8:30

(5 min averages)

Ambient CO and NOy correlate well during morning rush hour.

Can we apply the same technique toBeijing ambient measurements?

Nashville,Tennessee

1200

1000

800

600

400

200

0

CO

(ppb

v)

12:00 AM7/7/99

6:00 PM

150

100

50

0N

Oy (ppbv)

6:00 - 8:30

(5 min averages)



Nashville,Tennessee

4

3

2

1

0

CO

(ppm

v)

00:00 06:00 12:00 18:00 00:00

100

80

60

40

20

0

NO

y (ppbv)

Beijing August 2005Min Shao

Peking University

Compared to U.S. cities: CO and NOy much more variable; regional sources important much less of rush hour peak; higher CO, lower NOy.

2500

2000

1500

1000

500

0

CO

(ppb

v)

200150100500NOy (ppbv)

Year Slope r2

1994 10.2 ± 1.5 0.97 1995 8.5 ± 1.3 0.90 1999 6.3 ± 0.9 0.95




2500

2000

1500

1000

500

0

CO

(ppb

v)

200150100500NOy (ppbv)

Year Slope r2

1994 10.2 ± 1.5 0.97 1995 8.5 ± 1.3 0.90 1999 6.3 ± 0.9 0.95




4

3

2

1

0C

O (p

pmv)

706050403020100 NOy (ppbv)

slopes averages: 14 ± 5; r= 0.75hourly pts: 35 ± 3; r= 0.67

Beijing August 2005Min Shao

Peking University

Compared to U.S. cities: Much poorer correlation; CO (NOx?) dominated by non-road sources; higher slopes.

NotDirectly

0.01

2

4

68

0.1

2

4

68

1

2

4

68

10

Ben

zene

(ppb

v)

4 6 80.1

2 4 6 81

2 4 6 810

2

Acetylene (ppbv)

71 U.S. cities 1984-882002 NE US Coast2004 NE US Coast

• Are from road transport

• Correlate very well in the atmosphere

• Ratio gives emission ratio, whichprovides a test of emissioninventories

Benzene & acetylene:

• React slowly in the atmosphere -ambient benzene to acetylene ratioreflects emission ratio

Assumptions:

Benzene / acetylene: Conclusions:

VOC Speciation in U.S. on-road vehicle emissions: Characterization through ambient measurements

Speciated VOCs are inaccurate byfactors of 3 or more, depending on theconstituents.

(Fortin et al., ES&T, 39, 1403-1408, 2005)

-2.0

-1.5

-1.0

-0.5

0.0

20052000199519901985198019751970

Mea

n M

olar

Rat

io

Ambient Measurements 71 U.S. Cities ROSE SCAB SOS SCAQS NARE ITCT NEAQS

TexAQS

Inventory total

0.2

0.1

0.3

0.4

0.5

0.7

1.0

1.5

Inventory on-road

Accurately captures magnitude andtemporal trend of VOC emissions.

Comparison of inventory emissions with ambient measurements indicatesMOBILE6 model:

Inventories of U.S. emissionsa of two VOC species.

a units: 105 moles hr-1

Species NAPAP 1985

NEI 1996

NEI 1999

Inventory Benzene 10.69 4.08 3.98

Total Acetylene 9.24 4.53 5.38

Ratio 1.16 0.90 0.74

On-Road Benzene ------ 0.94 1.00

Acetylene ------ 1.74 3.15

Ratio ------ 0.54 0.32

Benzene/Acetylene Ratio


Figure 4. Emission ratios of various VOCs to CO measured in the Los Angeles Basin by the NOAA WP-3D aircraft in May, 2002 compared to the ratios calculated from SAPRC (Warneke et al., “Determination of Urban VOC Emissions Ratios and Comparison with Inventories”, in review, J. Geophys. Res.,2006)

16

14

12

10

8

6

4

2

086420

SAPRC VOC/CO (pptv/ppbv)

ALKANE1

ETHYLENE

ACETALDEHYDE

ACETONE

METHANOL

ACETYLENE

BUTANES

PENTANES

TOLUENE

PROPANE

BENZENEPROPYLENE

1:1

Comparison of inventory emissions with ambient measurements: Some successes, some problems.


0.01

2

4

68

0.1

2

4

68

1

2

4

68

10

Ben

zene

(ppb

v)

4 6 80.1

2 4 6 81

2 4 6 810

2

Acetylene (ppbv)

71 U.S. cities 1984-882002 NE US Coast2004 NE US Coast






Assumptions:



0.01

2

4

68

0.1

2

4

68

1

2

4

68

10

Ben

zene

(ppb

v)

4 6 80.1

2 4 6 81

2 4 6 810

2

Acetylene (ppbv)

2005 Beijing71 U.S. cities 1984-882002 NE US Coast2004 NE US Coast






Assumptions:



Beijing VOC levels and mixture is very much like U.S. cities in the 1980s

2005 Beijing data from Min Shao, Bill Kuster: Peking University

2

4

6

81

2

4

6

810

2

4

6

Ethy

lene

(ppb

v)

0.12 4 6 8

12 4 6 8

102

Toluene (ppbv)


• Correlate very well in the urban atmosphere

• Urban ratio gives emission ratio, which provides a test of emission inventories

• Only Benzene/Acetylene ratio has changed significantly in U.S.

Conclusions:




0.001

0.01

0.1

1

10

100

i-But

ane

(ppb

v)

0.001 0.01 0.1 1 10 100n-Butane (ppbv)



• Urban ratio gives emission ratio, which provides a test of emission inventories

• Only Benzene/Acetylene ratio has changed significantly in U.S.

Conclusions:




MOBILE6 has significant short-comings as well as strengths; for the U.S.:

• 1994 VOC emissions were accurate while CO emissions were overestimated by about a factor of two.

• Temporal trends of CO and VOC emissions are similar and accurate.• NOx emissions increased, not decreased, through the 1990’s. For 2006,

NOx emissions may be significantly underestimated.


Ambient measurements provide useful tests of VOC speciation in on-road emissions

Proposed Conclusion and Recommendation:It is not possible to accurately estimate emissions without input

from direct measurements (e.g. CEMS on point sources, satellite measurements, top-down tests through ambient measurements, inverse modeling, etc.)

An iterative process is required between inventory developers and measurers to reach accurate inventories.

What about Texas?

CO/NOx emission ratio inon-road mobile emissionshave decreased dramatically.

7.9 ± 0.1

7.3 ± 0.6

7.8 ± 0.7

9.1 ± 0.5

12.4 ± 0.7

Ratio in 2000

0.97-6.6 ± 0.3AIRS

0.86-5.8 ± 2.2Dallas

0.76-4.9 ± 2.5Houston

0.94-5.8 ± 1.5San Antonio

0.94-6.9 ± 1.7El Paso


NEI 1999 emission ratios estimates are too high and are worse now than in 2000

Expanded axes, AIRS data as reference

70

60

50

40

30

20

10

0

emis

sion

s (to

ns/d

ay)

DieselGasoline

Fuel-based

EPA Fuel-based

Fuel-based

EPAEPA

CO/10 VOC NOx

VOC agree well.

CO ≈ 40% high

Compare with Fuel-based Inventory

Nashville 1995 Road Transport Sources

(Harley et al., J. Geophys. Res., 106, 3559-3567, 2001)

Fuel-based emissions agreewith ambient CO/NOx

NOx agree well, butnot partitioning

MOBILE 5B emissions vs.Fuel-based in 1995:

MOBILE 6 emissions would degrade CO comparison

Documents

A Critical Evaluation of U.S. On-road Vehicle Emissions in ... 06... · David Parrish - NOAA ESRL ... VOC Speciation in U.S. on-road vehicle emissions: Characterization through ambient