Georgia Institute of Technology Air Pollutant Transport, Control and Modeling Issues in the Eastern United States Ted Russell Air Resources Engineering

Georgia Institute of Technology

Air Pollutant Transport, Control and Modeling Issues in the

Eastern United States

Ted RussellAir Resources Engineering Center

Georgia Tech

August 26, 2003


Acknowledgements

• GIT Colleagues– T. Odman, J. Boylan, A. Hakami, M. Bergin, D. Cohan,

Y. Hu, A. Unal, D. Tian, M. Kahn, H. Park, A. Marmur, J. Wilkinson, M. Chang, et al.

• SAMI– Financial support and, particularly, intellectual

contributions and guidance

• RFF Colleagues– J-S. Shih and A. Krupnick

• NSF and EPA


Issues• Ozone modeling morphing in to multi-pollutant

models– Confidence for ozone (?)– How about other species?

• What are the major uncertainties, and what is being done?

• Long range and near-field impacts of air pollutants – Is it me or my next door neighbor, or the guy down the

street?

• Georgia EAC Modeling– Issues– Preliminary results

• Other bits of interest– Reactivity, VOC vs. NOx (it never dies)


AQM State of the Science: Where are We?

• “One atmosphere”/“3rd generation” urban-to-regional/continental models are at the forefront– Combined

gas/aerosol/deposition & nested/multiscale/adaptive (?)

– Some built in diagnostic features

• Process analysis• Sensitivity analysis

• Dominant PM models– REMSAD (simplified)– Models 3/CMAQ– CAMX-PM (less application for

PM)

Regional Multiscale Model


Grids

Nested

Multiscale

Adaptive


How Good Are They?

• All evidence suggests that they describe the processes most affecting the evolution of ozone and (if equipped) particulate matter (o.k., many components of PM) after pollutant emission

Science (chemistry/physics)

Mathematics

Computational implementation

Evaluation

Uncertainties

Limitations & uncertainties

Limitations

Uncertainties, lack of confidence

Holes lead to:

Application Uncertainties, poor results


Confidence?• Should go in as a skeptic, however• Current models rather accurately (not perfectly) simulate

ozone and (most) associated species– Multiple, very different areas:

• Urban (L.A., Mexico City, NY, Tokyo, …)• Regional (Eastern U.S., NW, Europe, Asia, …)

– Different meteorologies• Across seasons, high, mid, low ozone

• Explain observed trends– 10 year trend in NOx, VOC, CO and ozone in Switzerland

• Doesn’t come easy:– 60% emissions, 30% meteorology, 5% model development, 5%

model application/evaluation • Emissions still most uncertain

• Expect “rapid” advances in PM– Supersites, RPO modeling, EPA-funded research


Performance (Good)Surface O3 in faqs4

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

Observed

Sim

ua

lte

d

y=0.56*x+0.030R2=0.49

Surface Ozone (40ppb cutoff)

05

10152025303540

Episode Day

MN

E(%

)

faqs12

faqs4

35%

Ozone

Sulfate


Performance (Not so good)• PM Performance (Seignuer et

al., 2003)– Errors from recent studies

using CMAQ, REMSAD• Organic carbon: 50-140%

error• Nitrate: 50-2000% error

– Understand the reason for much of the error in nitrate

• Deposition, heterogeneous reaction

• Ammonia emissions still rather uncertain

– OC more difficult• Understand part (most?)• More complex mixture• Primary/precursor emissions

uncertain Organic Carbon


Long vs. Short Range Impacts

• What is the relative contribution of local vs. nearby vs. distant emission sources?

• Approach– Apply air quality model with sensitivity analysis to

develop a matrix of interstate and intrastate impacts of NOx and SO2 emissions on ozone and PM2.5, respectively.

• NOx emissions had minimal impact on PM2.5 because the aerosol is relatively acidic (for now)

• Periods– RFF

» May, 1995: typical levels of ozone and sulfate» July, 1995: High levels of sulfate and ozone

– Southern Appalachians Mountains Initiative: Year average

Sources: Shih et al., (2003) and Boylan et al., (2003)


J uly

12

July

14

Ground Level ElevatedExample: Impact of Michigan NOx emissions downwind:


8-hour Ozone change for 30% Elevated Source NOx Reduction


g/m3/1000 tons/dayMay-July episodes

Sulfate Sensitivity to SO2 Emissions


GR

SM

LIG

O

SH

RO

JO

KM

CO

HU

SIP

S

JE

FF

OT

RC

SH

EN

DO

SO

Ozone W126 to Elevated NOxOzone W126 to Ground NOx

Wet Sulfate to SO2

Aerosol Sulfate to SO2

-3.000

-2.000

-1.000

0.000

Se

ns

itiv

itie

s (

%)

Sensitivities to 10% Emission Reductions in TN

Ozone W126 to Elevated NOx Ozone W126 to Ground NOx

Wet Sulfate to SO2 Aerosol Sulfate to SO2


Long Range vs. Near Field

• While long-range transport is important, near field impacts are major– Both SO2/sulfate and NOx/ozone

• Interstate transport very episode specific– SAMI looked at annual average, even stronger

conclusion about importance of local (near-field) controls


Georgia EAC Modeling

• Cities/Episodes– Augusta, North Georgia– Primary: August 2000 (episode selection suggests

representative)– Coming: August 1999, July 2001

• Issues– Modification of inventory (NET96NEI99)– Default vertical diffusivity

• Performance– Ozone– Precursors

• Preliminary 2007


36-km

4-km

12-km


Problem 1: Minimum vertical eddy diffusivity

• CMAQ Defaul of 1.0 m2/s

Plot of Simulated and Observed Surface Ozone Concentrations in Columbus, GA, Using a minimum vertical eddy diffusivity of 1.0m2/s.


Figure 6.11 Midnight Surface Ozone Concentrations on August 17th, 2000 in the FAQS 12-km Grid Using the Minimum Vertical Eddy Diffusivity of 1.0m2/s.

Still very high at midnight in most urban and suburban areas, over 30ppb and 60ppb


Simulated and Observed Surface NO Concentrations in the FAQS 12-km Grid for the Episode of August 11th - 20th,2000, the Minimum Vertical Eddy Diffusivity of 1.0m2/s was used in CMAQ.

Underestimate at night


Figure 6.14 Midnight Surface Ozone Concentrations on August 17th, 2000 in the FAQS 12-km Grid Using the Minimum Vertical Eddy Diffusivity of 10 -4m2/s.

• Minimum diffusivity reset to 10-4 m2/s (McNider and Pielke, 1981)

Corrected !


Time Series Plot of Simulated and Observed Surface Ozone Concentrations in Columbus, GA,the Minimum Vertical Eddy Difusivity of 10-4m2/s was used in CMAQ.


Figure 6.17 Simulated and Observed Surface NO Concentrations in the FAQS 12-km Grid for the Episode of August 11th - 20th,2000, the Minimum Vertical Eddy Difusivity of 10-4m2/s was used in CMAQ.

Corrected !


Late Afternoon Surface Ozone Concentrations on August 17th, 2000 in the FAQS 12-km Grid Using the Minimum Vertical Eddy Diffusivity of 10-4m2/s.

Isolated strange hot spots?


Time Series Plot of Simulated and Observed Surface Ozone Concentrations in Santa Rosa County, FL, the Minimum Vertical Eddy Difusivity of 10-4m2/s was used in CMAQ.

Extremely high ozone happened in the grid cells over mixed land-use where water is the majority


Problem 2: Atmosphere too stable over cells with mixed

water/other landuses• Stability determined by land-use with largest

fraction• Cell vertical diffusion went very stable over

lakes and coastal sites• Used 9-point averaging in such cells


Late Afternoon Surface Ozone Concentrations on August 17th, 2000 in the FAQS 12-km Grid Using the Minimum Vertical Eddy Diffusivity of 10-4m2/s and a 9-point averaging method.

• A 9-point averaging method was used to smooth …


Time Series Plot of Simulated and Observed Surface Ozone Concentrations in Santa Rosa County, FL, the Minimum Vertical Eddy Diffusivity of 10-4m2/s and a 9-point averaging method


Surface Ozone Daily MNB during the Episode of August 11th -20th, 2000 for the 106 Stations in the 12-km Grid or the 25 Stations in the 4-km Grid

Surface Ozone Daily MNE during the Episode of August 11th -20th, 2000 for the 106 Stations in the 12-km Grid or the 25 Stations in the 4-km Grid

Daily Bias and Errors of Ozone


Time Series Plot of Simulated and Observed Surface NO Concentrations at PAMS Station

NO

Diurnal Plots

Time Series Plot of Simulated and Observed Surface NOZ Concentrations at PAMS Station

NOy


Time Series Plot of Simulated and Observed Surface ARO1 Concentrations at PAMS Station

ARO1

Time Series Plot of Simulated and Observed Surface ALK1 Concentrations at PAMS

Station

ALK1


Estimated Change in Daily Maximum Ozone Concentrations in the 12km grid (on the left) and the 4-km grid (on the right) from August 17th 2000 to 2007 under the existed Federal Control

Strategies.

Difference of Daily Maximum Ozone: 2007-2000

Reductions typically 8-14 ppb


NOx vs. VOC: Did the Pendulum Swing Too Far?

• Are VOC controls beneficial, and where?

• Is attainment dependent upon NOx controls?

• What are the additional issues?

http://images.google.com/imgres?imgurl=www.sweepstakesbuilder.com/images/pendulum.jpg&imgrefurl=http://www.sweepstakesbuilder.com/wizard_overview.php&h=277&w=169&prev=/images%3Fq%3DPendulum%26start%3D20%26svnum%3D10%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26sa%3DN


Chemical RegimesRadical Limited: Abundant NO2 removes OH, inhibittingoxidation of VOCs and HO2/RO2 formation: Low utilization of NOx emissions

Volatile Organic Compounds (VOCs)

Nitr

oge

n O

xide

s (

NO

x)

Low O3

Transport

NOx Limited: Lack of NOx limits ozoneformation via photolysis, increased destruction of HO2/RO2: High utilizationof NOx emissions

High O3


Influence of Biogenics

Volatile Organic Compounds (VOCs)

Nitr

oge

n O

xide

s (

NO

x)

SEHot, sunny

SETypicalNE

Typical


Biogenic Emissions

(a)

(b)

(c)

Day 1 Day 2 Day 3

Biogenic emissions vary significantly by location and day

Source: Hanna et al., 2003


VOC SensitivityAnthropogenic VOC Isoprene


VOCVOCNOXNOXM

ob

ilM

ob

ilee

No

n-

No

n-

Mo

bile

Mo

bile

Georgia:

Even in Atlanta, VOC controls can be effective & some local NOx inhibition (seen in observations, too)


VOC vs. NOx

• VOC controls effective for controlling ozone, even on high ozone days– Can be relatively more effective on cooler, lower

ozone days– Additional benefits: lower carbonaceous PM, toxics

• NOx controls key to attainment in high biogenic areas

• Controlling VOC reactivity may be even more cost-effective– Coming ANPR (California leading)


Organic Reactivity Assessment

• Organics behave differently – need a way to quantify the impact each has on the ozone production.– Impact on ozone per mass of VOC emitted can vary by orders of

magnitude.• More cost effective management possible if one accounts for ozone

formation from emissions, not just VOC mass – Allows manufacturers flexibility in formulation– Controls perverse choices

• Less, but much more reactive VOC

• Pioneered in California for fuels– Moved to consumer products– Recent studies have shown applicability in eastern US, EPA moving forward

• Conducted large scale (LACAeastern US) 3-D simulations to assess reactivity– Variability, scales, approaches– Applied URM using SAPRC99.


Reactivity of organic compounds

• Reactivity of an organic compound, as a measure of its ozone formation potential can be defined as (Carter, 1994):

• By definition, reactivity assessment is sensitivity analysis.

• Assessed reactivities for various organics using URM & DDM-3D– Carter used CAMX and DDM

ii E

O

E

O

33


Relative Reactivity Metrics, MIR-3D

MIR-3D-1hr

-1.0

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

2MBT

BUTD

PRPE

ISOP

XYLM

ETHE

HCHO

ACRO

CCHO

APIN

TOLU

MCPT

ETOH

N_C5

N_C4

C6H6

MTBE

MEOH

IPOH

DODC

C2H6

BALD

Jul-95

Jul-10

May-95

May-10

Box mode MIR

Rel

ativ

e R

eact

ivit

y


SOS

• SOS activities moving towards Texas– Many SOS-affiliated teams applying expertise in the

Houston area– UTexas taking lead role

• State-of-Science document underway• No new EPA money for traditional SOS

– Significant “individual” investigator work– Funding for Texas studies


Summary• PM models coming of age

– Uncertainties, particularly in emissions, organics

• While interstate transport can be significant…– Local impacts typically large(st)

• Preliminary EAC modeling for Georgia suggests a 8-14 ppb reduction in ozone regionally– For August 2000 episode– 2001, 1999 episodes underway

• Reactivity appears to work in the East– Some differences than in the West

• VOC controls effective in cities, if not beyond– Even in Atlanta with all of its isoprene– NOx scavenging found in Atlanta as well

• Observed and modeled


FAQS Episodes

• August 11, 2000 – August 20, 2000– Primary FAQS period– High ozone statewide– FAQS measurements, Houston Supersite period– Col 8-hr, Mac. 8-hr, Atl&Mac&Col&Aug 8-hr, All 8-hr, Atl. 1-hr

• August 10, 1999 – August 21, 1999– High ozone and PM– Atlanta Supersite period– Mac 8-hr, Atl&Mac&Col&Aug 8-hr (part), All 8-hr (part), Aug

8-hr• July 5, 2001 – July 20, 2001

– High ozone and PM– FAQS measurements, ESP01 period– Atl 8-hr, Atl 1&8-hr, Atl&Mac&Col&Aug 8-hr, All 8-hr, Aug 8-

hr

Documents

Georgia Institute of Technology Air Pollutant Transport, Control and Modeling Issues in the Eastern United States Ted Russell Air Resources Engineering