David Parrish - NOAA Aeronomy Laboratory

• Review “Section 6.4: Evaluation of the strengths and weaknesses of current inventories and models.”

• Please feel free to provide critiques and feedback!!!

Today:

Preview of NARSTO Emission Inventory Assessment

EMIS Meeting, Portsmouth NH, August 3, 2004

1.0 Introduction

2.0 Overview of Emissions Inventories Provide an emissions inventory primer for decision makers

3.0 Vision for Future North American Emission Inventory Programs Set the target: define where we need to be in future years.

4.0 North American Emission Inventories List what currently exists

5.0 Inventory Tools Describe how are inventories constructed

6.0 Strengths and Weaknesses of Current Inventories and Models: Sensitivities, Uncertainties, and Applicability Describe how to evaluate Provide an evaluation of important sectors

7.0 Evolving Tools for Determining Emissions Describe advanced methods and methodologies

8.0 Roadmap for Future Emission Inventory Application and Development

NARSTO Emission Inventory Assessment

Emission Inventory Development

Emission Factors and

Models

Databases for Source Activity Levels

Defaults for Emissions Related

Variables

Point and Non Point

Source Data

Growth Factors

Starting Point for EI

Factor and Model Improvements

Regional/State/Local/Tribe/Provincial Improvements

EI Applications

Forecasting

Measurement Methods

Field Testing and Sampling

Local Activity Levels & Variables

Preliminary EI for Base

20xx

States and Other

Stakeholders

Improved EI for Base Year

20xx

Emissions Processor

Speciation Factors

Temporal Factors

Spatial Factors

Tests, Evaluations and Reviews

Emission Inventory Development

Emission Factors and

Models

Databases for Source Activity Levels

Defaults for Emissions Related

Variables

Point and Non Point

Source Data

Growth Factors

Starting Point for EI

Factor and Model Improvements

Regional/State/Local/Tribe/Provincial Improvements

EI Applications

Forecasting

Measurement Methods

Field Testing and Sampling

Local Activity Levels & Variables

Preliminary EI for Base

20xx

States and Other

Stakeholders

Improved EI for Base Year

20xx

Emissions Processor

Speciation Factors

Temporal Factors

Spatial Factors

Tests, Evaluations and Reviews

Evaluations of Road Transport Emissions:

• U.S. EPA emissions estimates have changed drastically. Danger Sign!!

• Most recent emissions estimates are uncertain by at least ~10% for VOC and NOx and ~50% for CO.

12

10

8

6

4

2

0200019951990198519801975

1989 NAPAP 1990 Trends report 1995 Trends report 2000 Trends report 2003 Trends report

14

12

10

8

6

4

2

0

140

120

100

80

60

40

20

0200019951990198519801975

2000 = MOBILE 52003 = MOBILE 6

• For 1985 the 2003 estimate is higher than previous estimates by factors of up to 1.35, 1.6 and 2.5 for NOx, VOC, and CO

History of temporal trends of road transport emissions in NEI

4

5

6

7

8

9

100

CO emissions (10

6

tons/year)

200019951990198519801975

3

4

5

6

7

8

9

10

Ambient CO levels (ppmv)

U.S. Mean

Ambient

2000 Trends

Report

2003 Trends

Report

• Rural U.S. CO levels have also decreased, but at a rate of only about 3%/yr. [Hallock-Waters et al., GRL, 26, 2861-2864, 1999]

Evaluations of Road Transport Emissions:

Temporal trend of CO emissions in NEI compared to ambient levels

• Trends in 2003 Report agree well with trend of ambient levels, but not 2000 Repoprt.

• Exception is latest 2 years.

• No information regarding the accuracy of the absolute emissions.

2000 = MOBILE 52003 = MOBILE 6

Urban

- 4.2 %/yr

- 4.4 %/yr

- 2.7 %/yr

Evaluations of Road Transport Emissions:

Temporal trend of CO/NOx ratio in NEI compared to ambient ratios

2000199519901985

Nashville, TNSummer

-8.8 ± 1.0 %/yr

Boulder, COWinter

-6.7 ± 0.5 %/yr

L.A., CA

2003 Trends Report-3.8 %/yr

2000 Trends Report-3.5 %/yr

10

7

5

15

20

25

Neither Trends Report matches recent low observed CO/NOx ratios, and 2003 is worse than 2000. Disagreement a factor > 2 by 2000.

(Parrish et al., JGR, 107, 4140, doi:10.1029/2001JD000720, 2002)

Neither Trends Report captures rapid decrease in CO/NOx ratios.

Current U.S. emissions estimates are too high for CO, and/or too low for NOx.

Evaluations of Road Transport Emissions:

Comparison of fuel-based and mileage-based emissions estimates

(Harley et al., JGR, 106, 3559-3567, 2001)

70

60

50

40

30

20

10

0

emissions (tons/day)

Diesel

Gasoline

Fuel-

based

EPA Fuel-

based

Fuel-

based

EPAEPA

CO/10VOC NOx

VOC agree well.

CO ≈ 40% high

Nashville 1995 Road Transport SourcesFuel-based emissions agree with ambient CO/NOx

NOx agree well, but not partitioning

EPA equivalent to 2000 Trends Report

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

MOBILE 6 emissions (2003 Trends Report) would degrade CO comparison

Evaluations of Road Transport Emissions:

Reconciliation of estimated emissions with ambient measurements

140

120

100

80

60

40

20

0

CO emissions

200019951990198519801975

12

10

8

6

4

2

0

NO

x

emissions

2000 Trends report

2003 Trends report

Inferred Emissions

CO emissions trend assumed to be - 5.1 %/yr (1990-2000 temporal trend of ambient CO levels).

CO to NOx emission ratio trend assumed to be - 8.8 %/yr (temporal trend for L.A. Nashville).

1995 NOx emissions from the 2000 Trends Report

are assumed to be accurate.

Assumptions:

CO emissions in the 2003 Trends Report are over estimated by a factor of 2 for 1990-2001 (worse lately).

Conclusions:

NOx emissions are within 20% of the 2003 Trends Report values for all years after 1993, but are increasing rather than decreasing.

Interestingly, the inferred NOx temporal trend closely follows the increasing trend of the first 5 years of the 1990s in the 2000 Trends Report.

Evaluations of Road Transport Emissions:

Reconciliation of estimated emissions with ambient measurements

140

120

100

80

60

40

20

0

CO emissions

200019951990198519801975

12

10

8

6

4

2

0

NO

x

emissions

2000 Trends report

2003 Trends report

Inferred Emissions

No substantial inconsistencies have been identified in the VOC emissions.

Conclusions:

VOC emissions trend for 1990-2000 (2003 Trends Report) is - 5.7 %/yr, very close to - 5.1 %/yr trend of ambient CO levels.

14

12

10

8

6

4

2

0200019951990198519801975

VOC emissions do not increase recently as CO.

Evaluations of Road Transport Emissions:

Evaluation of VOC speciation

• are from road transport

Conclusions: Benzene / acetylene ratio:

• exhibits long-term trends in response to VOC emission control measures

-2.0

-1.5

-1.0

-0.5

0.0

20052000199519901985198019751970

0.1

2

3

4

5

6

7

8

9

1

Mean Ratio

Ambient Measurements

71 U.S. Cities ROSE

SCAB SOS

SCAQS NARE

ITCT NEAQS

TexAQS

Inventory

Mea

n B

enze

ne/A

cety

lene

Rat

io

Assumptions: Benzene & acetylene:

• react slowly in the atmosphere - ambient benzene to acetylene ratio reflects emission ratio

• The inventory values are a factor of 3 to 4 higher than ambient, and trends do not agree.

• There is a critical need for a re- evaluation of the VOC speciation in the NEI.

Evaluations of Power Plant Emissions:

Nearly 50 plume studies on over 30 CEMS-equipped power plants (1995, 1997, 1999, 2000, and 2002).

Flux ratios agree on average within estimated uncertainty of ambient determinations (± 10%).

There are occasional significant discrepancies.

3x10-3

210NOy / CO2 emission ratio tabulated

2.0

1.5

1.0

0.5

0.0

emission ratio tabulated / correlation slope

10x10-3

86420SO2 / CO2 emission ratio tabulated

NOy = 1.10 ± 0.17SO2 = 1.04 ± 0.24

12x1027

86420

CO2 emission (molec/s)

2.0

1.5

1.0

0.5

0.0

emission tabulated / flux measured

80x1024

6040200

NOy, SO2 emission (molec/s)

NOy = 1.10 ± 0.31SO2 = 1.07 ± 0.25CO2 = 0.98 ± 0.29

Absolute NOx, SO2 and CO2 fluxes agree with CEMS data within the estimated uncertainty (±20% for optimum conditions).

CEMS data are accurate

Conclusions:

CEMS data are accurately integrated into the NEI

CO emissions can be more than a factor 10 higher than inventoried.

Evaluations of Power Plant Emissions:

emis

sion

tabu

late

d / f

lux

mea

sure

d

There are occasional significant discrepancies.

25

20

15

10

5

0NOx

,

ethene,

and

propene,

ppbv

12:52 PM8/27/00

12:56 PM 1:00 PM 1:04 PM

70

60

50

40

30

20

10

Ozone, ppbv

Sweeny FreeportChocolateBayou

Ethene

Propene

NOxavg

Isobutane

Inventory alkene emissions are low by the following factors:Sweeny Freeport Choc. Bayou

Ethene: 72 60 25 Propene 67 70 40

Coincident, highly elevated levels of NOx and reactive VOCs lead to rapid formation of high O3 levels. Models using inventoried emissions of VOCs fail.

Evaluation of Emissions from Texas Petrochemical Facilities

(Ryerson et.al., JGR, 108, 4249, doi:10.1029/2002JD003070, 2003)

PM2.5 source apportionment from chemical mass balance compared to the emission inventory in Denver, CO

Showed nearly twice the fractional contribution from fugitive dust emissions. Did not include cold start gasoline vehicle exhaust, which makes a substantial contribution in the CMB analysis. Underrepresented high emitter (i.e. poorly-maintained) gasoline vehicles.

Compared to emission inventory, CMB showed:

Inverse modeling application to

ammonia emission inventory

Showed 40% lower average annual emissions with strong seasonal cycle. Dramatic improvement in aerosol nitrate simulation (not included in inverse optimization)

Compared to emission inventory, inverse modeling showed:

Conclusions

Recommendations

Challenge inventories with ambient measurements.

Increase interaction and feedback between inventory and measurement folks.

(Appropriate sites, measurement techniques, data analysis.)

“Top down” tests do provide valuable evaluations of current inventories.

It is not yet possible to develop “bottom up” inventories that are accurate enough for many scientific uses without support of direct measurements (e.g. CEMS)

Still, “bottom up” inventories are indispensable

Global CO Emissions Inventory Enigmas:

Compared to U.S., Western Europe emits factor of 2 less with 40% more population. Lifestyle!

Western Europe emissions decreasing significantly while U.S. emissions nearly constant?

Road transportation emissions in U.S.: ≈ 2/3 of total. Really increasing?

4

3

2

1

020001995199019851980

U.S. Western Europe

Total

RoadTransport

+0.5 %/year(1990-1995)

-1.8 %/year(1980-1995)

+2.0 %/year(1990-1995)

+0.2 %/year(1990-1995)

-2.3 %/year(1990-1995)

+0.2 %/year(1980-1995)

Total

RoadTransport

EDGAR 2.0

EDGAR 3.2CO Emissions

“Everyone knows” that CO emissions have decreased dramatically over the last couple of decades

“National Vehicle Emissions Policies and Practices and Declining US Carbon Monoxide-Related Mortality”, J.A. Mott et al., J. Amer. Med.

Assoc., 288, 988-995, 2002. ≈ 12,000 deaths avoided in U.S.

≈ 86% increase in vehicle miles traveled 1975-1998

Death rates from CO Poisoning

140

120

100

80

60

40

20

0200019951990198519801975

NAPAP (1989)1990 Trends report1995 Trends report2000 Trends report2003 Trends report

U.S. National CO Emissions Inventory Enigmas:

U.S. EPA emissions estimates have changed drastically. Danger Sign!!

Looking at history of current estimates gives impression of increasing emissions.

(Road Transport Sources)

140

120

100

80

60

40

20

0200019951990198519801975

NAPAP (1989)1990 Trends report1995 Trends report2000 Trends report2003 Trends report

U.S. National CO Emissions Inventory Enigmas:

U.S. EPA emissions estimates have changed drastically. Danger Sign!!

Looking at history of current estimates gives impression of increasing emissions.

EDGAR

EDGAR inventory may be affected by this changing of models (Road Transport Sources)

CO to NOx Ratio in Urban Vehicle Exhaust

1200

1000

800

600

400

200

012:00 AM7/7/99

6:00 PM

150

100

50

0

6:00 - 8:30

(5 min averages)

Ambient CO and NOy correlate well during morning rush hour.

Ambient measurements characterize CO/NOy emission ratio in vehicle exhaust.

CO to NOx automotive exhaust

emission ratios

3000

2000

1000

0 16012080400NOy (ppbv)

Year Slope r2 1989 18.9 ± 2.70.95 1991 15.7 ± 2.30.94 1996 12.7 ± 2.00.87 1998 8.9 ± 1.30.94

6:00 - 10:00Boulder, CO(5 min averages)

2500

2000

1500

1000

500

0 200150100500NOy (ppbv)

Year Slope r2

1994 10.2 ± 1.50.97 1995 8.5 ± 1.30.90 1999 6.3 ± 0.90.95

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

• Ratio of CO to NOx in vehicle exhaust has decreased dramatically at these two sites from 1989 to 1999.

2000199619921988

Nashville-8.8% / yr

Boulder-6.7% / yr

U.S. Inventory-3.4% / yr

L.A. Boulder-DenverInventory-7.1% / yr

5

7

10

15

20

Comparison to Emission Inventories

• VOC emissions parallel CO emissions.

• The character of urban photochemistry must have changed as well.

• National inventory does not accurately reflect observed decrease.

2000199619921988

Nashville-8.8% / yr

Boulder-6.7% / yr

U.S. Inventory-3.4% / yr

L.A. Boulder-DenverInventory-7.1% / yr

5

7

10

15

20

(National Air Pollutant Emission Trends, 2000)

• Colorado inventory does reflect Boulder decrease.

2000199619921988

Nashville-8.8% / yr

Boulder-6.7% / yr

U.S. Inventory-3.4% / yr

L.A. Boulder-DenverInventory-7.1% / yr

5

7

10

15

20

(Colorado Department of Public Health and Environment)

CO

to

NO

x R

atio

Emissions ratios: from near-field aircraft transects in mixed layer

32.0

31.9

31.8

31.7

31.6

31.5

31.4

Latitude

-96.6 -96.5 -96.4 -96.3 -96.2

Longitude

wind

Limestone PP

10

8

6

4

2

0

NOy, ppbv

TX

385

380

375

370

365

CO2

, ppmv

5:54:00 PM9/10/00

5:55:00 PM 5:56:00 PM

Time, GMT

25

20

15

10

5

0

NOy

and SO2, ppbv

CO2SO2NOy

Emissions ratios: from near-field aircraft transects in mixed layer

32.0

31.9

31.8

31.7

31.6

31.5

31.4

Latitude

-96.6 -96.5 -96.4 -96.3 -96.2

Longitude

wind

Limestone PP

10

8

6

4

2

0

NOy, ppbv

TX

40

30

20

10

0

SO2, ppbv

384382380378376374372370

CO2, ppmv

16

14

12

10

8

6

4

2

0

NOy, ppbv

Emissions ratios: from near-field aircraft transects in mixed layer

32.0

31.9

31.8

31.7

31.6

31.5

31.4

Latitude

-96.6 -96.5 -96.4 -96.3 -96.2

Longitude

wind

Limestone PP

10

8

6

4

2

0

NOy, ppbv

TX

40

30

20

10

0

SO2, ppbv

384382380378376374372370

CO2, ppmv

16

14

12

10

8

6

4

2

0

NOy, ppbv

Obs. CEMS NOx/CO2 1.2±0.1 1.4±0.1 SO2/CO2 1.8±0.2 1.9±0.2

Emissions ratios: from near-field aircraft transects in mixed layerslopes from multiple transects vary by ±2%combined instrumental uncertainties typ. ± 10%

provides an exacting test of inventory ratios

32.0

31.9

31.8

31.7

31.6

31.5

31.4

Latitude

-96.6 -96.5 -96.4 -96.3 -96.2

Longitude

wind

Limestone PP

10

8

6

4

2

0

NOy, ppbv

TX

40

30

20

10

0

SO2, ppbv

384382380378376374372370

CO2, ppmv

16

14

12

10

8

6

4

2

0

NOy, ppbv

Obs. CEMS NOx/CO2 1.2±0.1 1.4±0.1 SO2/CO2 1.8±0.2 1.9±0.2

Absolute emission rates: calculate flux across plume transect

White et al., Science, 194, 187-189, 1976.

Absolute emission rates: calculate flux across plume transect365

360

355

350

34520:287/7/99

20:29 20:30

20

10

0

60

50

40

30

20

10

0

Species Calculated Tabulated Flux Emissions

NOy 1.27 x 1025 1.24 x 1025

SO2 6.0 x 1024 5.6 x 1024

CO2 4.0 x 1027 4.3 x 1027

units: molec / s

Thomas Hill plantRandolph County, MO

July 7, 1999

20 km/1.5 hrs downwind

Estimated uncertainty of±20% in flux calculationunder optimal conditions

Application to CO emissions from power plants:

400

350

300

250

200

150

100

806040200

NOy, ppbv

440420400380360

CO2, ppmv

D

9/3/009/13/00

400

350

300

250

200

150

100

806040200

NOy, ppbv

440420400380360

CO2, ppmv

C

9/3/009/13/00

Power plant combustion is carefully monitored for maximum efficiency;most power plants emit undetectable amounts of CO, most of the time

Martin Lake (TX) Monticello (TX)

Three out of four lignite-coal-fueled power plants showed substantiallyenhanced CO emissions, up to factors of 30 greater than inventory values;also observed in bituminous- and gas-fired electric utility power plants.

J. Environ. Monit., 5, 35-39 (2003)

expectedslopes, frominventoryratios

Application to petrochemical industrial emissions: Houston, TX

Compare measured (alkene/NOx) to inventory (alkene/NOx) • Electra NOx and VOC measurements accurate to ±10%• 2000 PSDB inventory NOx data are accurate to ± 30%• any larger discrepancies in (alkene/NOx) comparison attributed to errors in the alkene emissions inventory

25

20

15

10

5

0NOx

,

ethene,

and

propene,

ppbv

12:52 PM8/27/00

12:56 PM 1:00 PM 1:04 PM

70

60

50

40

30

20

10

Ozone, ppbv

Sweeny FreeportChocolateBayou

Ethene

Propene

NOxavg

Isobutane

Application to petrochemical industrial emissions: Houston, TX

25

20

15

10

5

0NOx

,

ethene,

and

propene,

ppbv

12:52 PM8/27/00

12:56 PM 1:00 PM 1:04 PM

70

60

50

40

30

20

10

Ozone, ppbv

Sweeny FreeportChocolateBayou

Ethene

Propene

NOxavg

Isobutane

Inventory alkene emissions are low by the following factors:Sweeny Freeport Choc. Bayou

Ethene: 72 60 25Propene 67 70 40 (JGR, 108(D8), 4249, doi:10.1029/2002JD003070, 2003)

Nashville and Boulder are representative

199819961994199219903

4

5

6

7

8

Boulder-6.3 ± 1.5% / yr

r2 = 0.89

Nashville-4.8 ± 2.4% / yr

r2 = 0.63

U.S. medianambient levels-5.2 ± 0.8% / yr

r2 = 0.95