1 of 24

GEOS-Chem Users Meeting April 12, 2007

U.S. INFLUENCE ON TROPOSPHERIC OZONE AND THE EFFECTS OF RECENT EMISSION REDUCTIONS

Modeling: Rynda Hudman*, Lee Murray, Daniel Jacob, Solene Turquety, Dylan Millet, Shiliang Wu

NEI 99 Emissions: Alice Gilliland

ICARTT Observations: Melody Avery, Tim Bertram, Ron Cohen, Jack Dibb, Frank Flocke, Allen Goldstein, John Holloway, Andy Neuman, Tom Ryerson, Glen Sachse, Hanwant Singh, P J Wooldridge

ICARTT Flight Tracks

2 of 24

North America

🡪 Europe

NO_x stationary sources 50%

Anthropogenic CO 60%

4

2

6

8

Alt (km)

10

O₃(ppbv)

BB NA FF Lightning

NO_x/flash

4X larger than previously assumed!

Export well constrained

🡪 effects on O₃ & OPE

Northern Hemisphere Burden

ICARTT CONSTRAINTS ON SUMMERTIME EMISSIONS & HEMISPHERIC OZONE

Before I dive in, I want to summarize the major findings of my work and then go into detail on each one.

First for the inflow:

Asian pollution plumes are uplifted over the continent in cyclones in the spring, are transported quickly at 5-8 km. Some continue over the U.S. but others subside. We find that during subsidence PAN decomposition to form NOx is a major and possibly dominant component of ozone enhancement in these plumes. Strong dilution takes place as plumes entrain into the boundary layer reducing the impact on U.S. surface sites. California mountain sites, however, are more sensitive to Asian pollution because of their contact with the free troposphere.

Over the continent we used ICARTT observations to confirm decreases in stationary NOx sources over the U.S. from the NOx SIP Call which amounts to a 22% decrease in NO_x emissions during the summer. We then look at resulting ozone decreases in terms of regional ozone production efficiency.

We find the anthropogenic CO source over the U.S. is over estimated by 60% making it less than the biogenic source in the summer. We find that the NOx oxidation in the BL and export to the free troposphere is well understood. At 3-4 km HNO3 dominates the export but at higher elevations PAN is dominant supporting evidence of a large source of ozone through PAN export.

We constrain NOx by lightning. It is a factor of 4 higher than what is previously thought and has a 10-15 ppbv impact on ozone. We then use these constraints to get estimates on enhancements to the ozone over the U.S. and downwind. We find equal enhancements from lightning and FF in the upper troposphere.

3 of 24

May-August 2004 NA Fire Inventory

[Turquety et. al, 2007]

EPA National Emissions Inventory 1999 v1 (w/ modifications to VOCs)

Power plant and Industry NOx 50%

Anthropogenic CO 60%

GEOS-CHEM SIMULATION

(v7.02.04 w/ modifications below)

NOx Lightning Emissions

Lightning X4 over U.S. & distributed to tropopause

[Price and Rind, 1992]

Modifications from ICARTT constraints in blue (improved)

Flash counts (flashes/km²/s)

CO emissions

NO_xemission

4 of 24

GEOS-CHEM VS. ICARTT Mean comparison along the flight tracks

Large UT NO_xbias

BL bias in CO and NO_x

Ozone FT bias 5-10 ppbv

Measurements (WP-3D, DC-8): CO (J. Holloway, G. Sachse), NO_x (T. Ryerson, R. Cohen, W. Brune), PAN (F. Flocke, H. Singh), HNO₃(A. Neuman, J. Dibb), ozone (T. Ryerson, M. Avery)

Using this simulation GEOS-Chem simulation sampled along the NASA DC-8 and NOAA WP3 flight tracks, we found some discrepancies in our current understanding of the North American NOy budget, particularly in upper tropospheric NOx and ozone concentration as well as in changes in emissions over the U.S. since 1999. Shown here in red are mean simulated profiles using our standard simulation. The black are mean observations. The WP3 and DC-8, shows a CO bias in the lowest two km, which Lee Murray discussed in his presentation yesterday. For NOx we also see a model high bias in the lowest two km. Striking are two features in the upper troposphere…elevated NOx underestimated in the model by a factor of 4 with current emissions estimates and a 5-10 ppbv bias in oxone throughout the free tropospheric over the U.S. with current emissions.

5 of 24

Observed Simulated Improved Simulation

DC-8 Midwest Median NO₂

Median Model / Observed NO_x(0-2 km)

[ratio]

Large overestimate powerplant/industry dominated Midwest and in the South

50% reduction in power and industry source as determined by Frost et al., [2006] improves boundary layer NO₂ simulation

ICARTT OBSERVATIONS CONFIRM LARGE DECREASE SINCE 1999 IN INDUSTRY/POWER SOURCE

Measurements (WP-3D, DC-8): T. Ryerson (NO₂), Ron Cohen/Tim Bertram (NO₂)

NO₂ (ppbv)

6 of 24

OZONE REDUCTIONS RESULTING FROM DECREASE IN NO_x EMISSIONS

Can we see changes in OPE due NO_x emission reductions in dO₃/dCO in U.S. outflow?

Requires good estimate of CO source…..

Regional differences in Δozone, can be explained by OPE:

OPE

Midwest: 2.5-3.5

Southeast: 4-5.5

7 of 24

BOTH AIRCRAFT AND SURFACE DATA SUGGEST NEI 99 CO EMISSIONS ARE 2.5 TIMES TOO HIGH

Aircraft (0-1.5 km)

Chebogue Point (surface)

OBSERVED

SIMULATED (NEI99)

SIMULATED (anthro CO reduced by 60%)

Measurments:

J. Holloway, G. Sachse

Measurments:

A. Goldstein/ Dylan Millet

Hudman et al. [2007b]

8 of 24

SCATTERPLOT OF SIMULATED TO OBSERVED CO

Parrish [2006] finds on-road source overestimated by 50% in NEI 99 (🡪~33% reduction in NEI source due to this bias)

CO decrease trend 3.7% yr^-1 (1987-2002), (🡪 12% reduction in anthropogenic CO source since 1999 due to trend)

🡪 From these estimates 2004 emissions 45% lower than NEI 99 (Our estimate 60%)

9 of 24

ANTHROPOGENIC CO SOURCE IN THE UNITED STATES IN SUMMER IS NOW LOWER THAN BIOGENIC SOURCE

CO ANTHROPOGENIC EMISSION (11.5, 4.6)

CO SOURCE FROM ANTHROPOGENIC VOC OXIDATION (1.8, 1.8)

CO SOURCE FROM ISOPRENE OXIDATION (6.7, 6.7)

CO SOURCE FROM OTHER BIOGENIC OXIDATION (2.4, 2.4)

NEI 99

NEI 99 with 60% reduction in CO

CO SOURCE FROM OTHER BIOMASS BURNING OVER CONTINENTAL U.S.(0.16, 0.16) 🡨 Note: Fires in Canada and Alaska ~19 Tg CO

SOURCE TYPE (Tg CO)

10 of 24

OZONE-CO CORRELATIONS SHOW DECADAL INCREASE

Obs during the early 90s show dO₃/dCO ~ 0.3 – 0.4 [Chin et al., 1994; Parrish et al., 1998].

Change consistent with decadal changes in emissions and OPE over the Northeaster United States

Overestimate of tropical background

WINDS FROM W-SE

ALL WIND DIRECTIONS

Aircraft (0-1.5 km, 11-5pm LT)

Chebogue Point

Here I am showing ozone/CO correlations from the DC-8 aircraft and the Chebogue Point surface site on the southern tip of nova ScotiaThe Aircraft data are from 11 am – 5pm, excluding fresh plumes. For chebogue point we show data from the W-SE which sampled NA outflow. ThedO3/dCO in the obse is 0.47 (R2=0.54) and the corresponding model ratio is 0.36 (R2-0.21). This is caused by an overestimate of the ozone background in the Southeast and tropical airmasses. Looking at the regions geographically, we found what really dominates this slope is emission from the NE U.S.

The slope at Chebogue Point is lower 0.4, consistent although slightly lower than observations. The same problem exists with background as above. Previous studies found slopes of 0.3-0.4. The higher values could be due to decadal changes in U.S. CO and NOx emissions.

11 of 24

SUMMERTIME NORTH AMERICAN OZONE ENHANCEMENTS

Biomass Lightning Anthropogenic

Simulated Observed All

North American Source	NO_x Emission (Tg N)	Ozone Production Efficiency	Hemispheric ozone enhancement (Tg, %)
Lightning	0.28	32.5	9.1 (5.1%)
Biomass burning	0.32	17.5	5.6 (3.1%)
Fossil fuel	0.72	15	10.9 (6.1 %)
All	1.32	19	25.6 (14.3 %)

NA Enhancement to Hemispheric Ozone

ICARTT DC-8

~ Equal contributions for lightning and anthropogenic emissions in free troposphere and to NH burden

12 of 24

13 of 24

OBSERVED dO₃/dCO INCREASE OVER THE PAST DECADE CONSISTENT WITH UNDERSTANDING OF OPE AND SOURCES

dO₃/dCO ≅ OPE (dO₃/dNOx) * NO_x/CO source ratio (dNO_x/dCO)

Consider NE U.S.,

July 1 – August 15, 2004 (With ICARTT Constraints)

Anthro = 1.2 Tg CO, 0.10 Tg N Biogenic = 0.87 Tg CO

July 1 – August 1994

4.9% anthro decrease/year in urban air [Parrish, 2006] 🡪 Total CO 26% higher

22% stationary NO_x reduction [Hudman et al., 2007] 🡪 Anthro NO_x 15% higher

🡪 OPE lower by ~9%

🡪 NO_x/CO source ratio lower by ~19%

~28% increase in dO₃/dCO expected 0.3-0.4 (90s) 🡪 0.4-0.5 (present)

Multiply dO₃/dCO * ECO 🡪1.5 Gmol ozone d^-1(Ozone flux consistent with 1990s estimates)

14 of 24

Fuel - based inventory + CO & CO/NO_xambient trends 🡪 inferred emissions

[Parrish et al., 2006]

EVALUATION OF EPA NEI99 ON-ROAD TRANSPORT EMISSIONS

On-road emission estimates

*Note: 2004 trends report is same as NEI99

On-Road transport inventories (1995)

On-road CO emissions too high by 50%

15 of 24

UNITED STATES NO_x REGULATIONS AIMED AT REDUCING OZONE SMOG

1998 - The EPA NO_x SIP Call mandated that 22 eastern states & DC revise their SIPs to meet NO_x emissions reductions by 2003 (Phase 1) and further reduction by 2007 (Phase 2).

2003 - All 22 states reduced NO_x point source emissions to their Phase 1 levels.

2004 - Frost et al., [2006] determined that as a result emission rates of power plant NO_x have decreased approximately 50% between 1999 and 2003.

2005 - NOx levels in the U.S. will likely continue to drop. In March 2005, the EPA issued the Clean Air Interstate Rule, which will, permanently reduce NO_x emissions to 60% of 2003 levels in 25 eastern states, and the DC.

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

50% NOx PP reduction

Frost et al., [2006]

NOx SIP

SIP Phase 1

SIP Phase 2

CAIR

16 of 24

Frost et al., [2006]

17 of 24

CO OVERESTIMATE IS NOT CAUSED BY INSUFFICIENT VENTILATION OR BIOGENIC SOURCE

Boundary layer ventilation is constrained by vertical profile of short-lived VOCs

Biogenic VOCs are well-constrained by successful simulation of formaldehyde

Propane: E. Atlas CH₂O: Alan Fried

18 of 24

[Provided by D. Millet]

CHEBOGUE POINT CO TIMESERIES

Bias still exists during outflow events

7/1

8/1

19 of 24

UT NO_x OBSERVATIONS POINT TO A LARGER THAN EXPECTED LIGHTNING NO_x SOURCE

Hudman et al. [2007a]

GEOS-Chem (Lightning X4)

Observed

NO_x(8-12 km)

[ppbv]

DOESN’T APPEAR TO BE A NO_x LIFETIME ISSUE

NO: W. Brune, NO₂: R. Cohen/T Bertram

20 of 24

Lightning parameterization (flashes/km²/s):

Land : ~CTH^4.9 , Ocean: ~CTH^1.73

CTH= Cloud Top Height

Price and Rind [1992]

GEOS-Chem Vertical Distribution

GEOS-Chem

NLDN

[Flashes km² s]

FLASH RATES WELL SIMULATED POINTING TO A LARGER YIELD/FLASH AT NORTHERN MIDLATITUDES

Flash Comparison

Pickering et al., [1998]

21 of 24

[Huntrieser et al., 2005]

PEAK CURRENT AS A FUNCTION OF LATITUDE

22 of 24

[Ken Pickering]

NO PRODUCTION RATE CALCULATED FROMR RECENT CAMPAIGNS

Standard GEOS-Chem mean flash rate was 125 mol flash^-1 (Improved 🡪 X4 🡪 500 mol flash^-1)

23 of 24

OZONE COMPARISON INTEX-NA SOUTHEAST U.S. Increase in lightning yield X4 to 500 mol/flash has ~10 ppbv effect on ozone

NO₂

O₃

Hudman et al. [2007a]

…suggests great sensitivity of ozone to climate change

Observed Simulated Improved Simulation

2004 was not an anomalous lightning year

24 of 24

Hudman et al. [2007b]

NORTH AMERICAN ENHANCEMENT TO HEMISPHERIC OZONE

Here I am showing ozone enhancements from different North American NOx sources at different altitudes. Qinbin Li found a permanent upper level cyclone along the southern U.S. allowing ozone buildup. Cooper found and associated 10-17 ppbv from ightning. We find a similat 10-17 ppbv stretching woenwind across subtropical Atlantic to Europe.

Th emean NA anthro enhancement is at the surface but shifts NE in the Atlantic free troposphere reflecting frontal passages. NA emissions extend further north than lightning.

Biomass burning is highest in boreal regions reflecting the high fire ywar. Emissions extend to the Arctic where they dominate over anthro, lightning at the surface but not in the upper troposphere. Enhancements of 3-5 ppbv are seen over Western Europe similar to anthropogenic emissions.