Annual Simulations of Models-3/CMAQ: Issues and Lessons Learned Pat Dolwick, Carey Jang, Norm...

Annual Simulations of Models-3/CMAQ: Annual Simulations of Models-3/CMAQ:

Issues and Lessons LearnedIssues and Lessons Learned

Pat Dolwick, Carey Jang, Norm Possiel,

Brian Timin, Joe Tikvart

Air Quality Modeling Group

USEPA/OAQPS/EMAD

Research Triangle Park, North Carolina

M3 User Workshop, 10/21/2002

OUTLINEOUTLINE

Applications and evaluations of

Models-3/CMAQ at USEPA/OAQPS

Issues related to annual modeling of

ozone and PM

Preliminary comparison results between

CMAQ-2002 and -2001 releases

On-going and upcoming activities

Models-3/CMAQ Applications Models-3/CMAQ Applications and Evaluations at EPA/OAQPSand Evaluations at EPA/OAQPS

Western U.S. ApplicationWestern U.S. Application• Episodic O3, July 96, 36/12 km, evaluation, (AWMA 2001)

Annual Continental U.S. ApplicationAnnual Continental U.S. Application • 1-atmosphere, annual 1996, 36-km, evaluation &

diagnostics, on-going for 2001, (AWMA 2001 & 2002) Eastern U.S. ApplicationEastern U.S. Application

• 1-atmosphere, July 95, urban applications, 36/12/4-km, emissions control & growth (on-going)

Intercontinental Transport modelingIntercontinental Transport modeling• Trans-Oceanic transport of air pollutants and nested U.S. &

Asia/China modeling for year 2001

12 km western US ozone domain

36 km western US ozone domain

36 km eastern US domain

Models-3/CMAQ Modeling: Domain Maps

36 km Annual National US domain

12 km domain

4 km domain

Annual Continental U.S. ApplicationAnnual Continental U.S. Application

Features :Features :• Annual CMAQ modeling Annual CMAQ modeling over the continental U.S.,over the continental U.S., ““One-Atmosphere”One-Atmosphere”

Model Setup :Model Setup :• Annual PMAnnual PM and O and O3 3 (1996)(1996)• 36-km, 8/12 vertical layers36-km, 8/12 vertical layers• Meteorology : MM5Meteorology : MM5• Emissions Processing: SMOKEEmissions Processing: SMOKE• Model Evaluation: Compared against observed data Model Evaluation: Compared against observed data

(IMPROVE & CASTNET) & REMSAD(IMPROVE & CASTNET) & REMSAD• Papers published in AWMA (2001 & 2002)Papers published in AWMA (2001 & 2002)

July 1, 1996

NOx Emissions SO2 Emissions

Run Time & Disk Space Requirementfor Annual CMAQ Simulation

CMAQ annual run time was ~14 daysCMAQ annual run time was ~14 days• 4 CPUs (400 MHz) Sun Workstations (Enterprise 450); 4 CPUs (400 MHz) Sun Workstations (Enterprise 450);

Running 4 quarters of the year concurrentlyRunning 4 quarters of the year concurrently• ~4 hours/model day (8 layers)~4 hours/model day (8 layers)

Model-Ready Inputs/Outputs (entire year)Model-Ready Inputs/Outputs (entire year)• Emission (SMOKE/ECIP) : 40 GBEmission (SMOKE/ECIP) : 40 GB• Meteorology (MM5/MCIP) : 100 GBMeteorology (MM5/MCIP) : 100 GB• Outputs (M3/CMAQ) : 250 GB, compressed to 9 GBOutputs (M3/CMAQ) : 250 GB, compressed to 9 GB

The model runs >4 times faster on new Linux PCsThe model runs >4 times faster on new Linux PCs• 1.6GHz AMD Athlon MP (single CPU)1.6GHz AMD Athlon MP (single CPU)

Current modeling are transitioned to Linux PCsCurrent modeling are transitioned to Linux PCs (parallel processing on the PC/Linux cluster is (parallel processing on the PC/Linux cluster is being investigated)being investigated)

144144127127

194194

132132

126126119119

134134

1391397676

5656

107107

113113

149149 6262

137137

215215 131131

9999

Annual Continental U.S. CMAQ Modeling: O3

(July Max in ppb, 1996)

2626

1919

2626

2424

2424

2222

1616

22222424

2626

1414

1414

1010

1515

1818

991616

99

2323

1111

1212

Annual CMAQ Modeling: Visibility (January Average in Deciview)

Annual CMAQ Modeling: Wet Deposition

Sulfur Wet Deposition Nitrogen Wet Deposition

(January average)

Models-3/CMAQ Simulation: Annual Average

PM 2.5 Sulfate PM

Organic PMNitrate PM

Observation (IMPROVE)

Mo

del

Annual Average: Total PM 2.5 mass

Observation

Mo

del

NO3

SO4

OC

Annual Average: PM 2.5 Species (SO4, NO3, OC)

Models-3/CMAQ : Monthly Average (July)

PM 2.5 Sulfate PM

Organic PMNitrate PM

Observation

Mo

del

NO3

SO4

OC

Summer Average: PM 2.5 Species (SO4, NO3, OC)

Models-3/CMAQ : Monthly Average (January)

PM 2.5 Sulfate PM

Organic PMNitrate PM

Winter Average: PM 2.5 Species (SO4, NO3, OC)

Observation

Mo

del

NO3

SO4

OC

Annual CMAQ Modeling:

CB4 vs. RADM2

Nitrate PM (January avg, 1996)

Annual CMAQ Modeling:

CB4 vs. RADM2

Sulfate PM (July avg, 1996)

NH3 Sensitivity Modeling

BaseBase 50% NH 50% NH33 reduction reduction

1.161.16

0.620.62

1.241.24

0.530.53

0.700.70

0.080.081.691.690.090.09

Nitrate PMNitrate PM : : (January Avg, 1996)(January Avg, 1996)

On-going and Upcoming Modeling On-going and Upcoming Modeling Activities at OAQPS/AQMGActivities at OAQPS/AQMG

Continue model evaluation and sensitivity test for Continue model evaluation and sensitivity test for the annual 1996 continental US runthe annual 1996 continental US run

• Examine nitrate and organic PM issues

–Emissions issues (NH3, BEIS3, Fire, CA emis. etc.)

–Vertical layer sensitivity (8-14 layers)

–Nitrate nighttime chemistry issue– “N2O5 + H2O --> 2HNO3” and “NO2 + OH --> HNO3”

–Dry deposition Issue

–Vertical diffusion sensitivity

Modeling support for regulatory analysisModeling support for regulatory analysis Clear Skies Initiative (on-going) PM Transport Rule (2002/03) Nonroad Rule (2002/03) Section 812 CAAA Prospective Update (2003) 8-Hour Ozone Implementation Guidance

(2002/03) PM NAAQS Review (2003) Regional Haze Rule (on-going)

On-going and Upcoming Modeling On-going and Upcoming Modeling Activities at OAQPS/AQMG Activities at OAQPS/AQMG (contd.)(contd.)

Improvement of base model simulationImprovement of base model simulation• Increase vertical resolution (from 8 to 12 layers)• Monthly NH3 emissions adjustments from animal

husbandry & fertilizer operations (based on ORD inverse modeling)

• BEIS-3/BELD-3• NOx emissions from new Nonroad model• California NOx & VOC emission adjustments• Updated plume rise algorithm in SMOKE• Upgrade to latest Models-3/CMAQ release (July

2002)

On-going and Upcoming Modeling On-going and Upcoming Modeling Activities at OAQPS/AQMG Activities at OAQPS/AQMG (contd.)(contd.)

National 1996 CMAQ Modeling:

Peak Ozone (July 1996)

8-layer CMAQ 12-layer CMAQ

Old Base vs. New Base

Sulfate PM (Monthly average, July 1996)

8-layer CMAQ 12-layer CMAQ

Annual CMAQ Modeling:

Old Base vs. New Base

Nitrate PM (Monthly average, January 1996)

8-layer CMAQ 12-layer CMAQ

Annual CMAQ Modeling:

Old Base vs. New Base

AnthropogenicSOA

BiogenicSOA

(BEIS2 vs. BEIS3)

8-layer CMAQ 12-layer CMAQOrganic PM (July 1996)

Annual CMAQ Modeling:

Old Base vs. New Base

CMAQ Seasonal Ozone (8-hr max) Evaluation

Spring

Fall

Summer

WinterObservation

Mo

del

CMAQ Seasonal Sulfate PM Evaluation

Spring

Fall

Summer

WinterObservation

Mo

del

CMAQ Seasonal Organic PM Evaluation

Spring

Fall

Summer

WinterObservation

Mo

del

CMAQ Seasonal Nitrate PM Evaluation

Spring

Fall

Summer

WinterObservation

Mo

del

CMAQ Seasonal

Total NO3 (HNO3 + Nitrate PM)

Evaluation

Spring

Fall

Summer

WinterObservation

Mo

del

IMPROVE PM Species National East WestPM2.5 8% 3% 16%Sulfate Ion -2% 5% -19%Nitrate Ion 200% 262% 150%Elemental Carbon 14% -2% 28%Organic Aerosols 1% -30% 26%Soil/Other 80% 204% 43%

CMAQ vs. IMPROVEModel performance : Annual average

CB-IV SAPRC-99CB-IV SAPRC-99

((JanuaryJanuary Average) Average)

Nitrate Nighttime Chemistry Sensitivity“N2O5 HNO3” rxn rate (sensitivity test)

Base KvBase Kv(min) (min) = 5= 5 Kv(min) = 1, StableKv(min) = 1, Stable up to 3up to 3rdrd layer, ~230 m layer, ~230 m

((JanuaryJanuary Average) Average)

Kv (vertical diffusion) Sensitivity

((JanuaryJanuary 19, 2002) 19, 2002)

NH3 Dry Deposition Issue in CMAQ

NHNH33 Dry Dep. (V Dry Dep. (Vdd) Air Temp.) Air Temp.

Improvement of base model simulationImprovement of base model simulation• Increase vertical resolution (from 8 to 12 layers)• Monthly NH3 emissions adjustments from animal husbandry &

fertilizer operations (based on ORD inverse modeling)• BEIS-3/BELD-3• NOx emissions from new Nonroad model• California NOx & VOC emission adjustments• Updated plume rise algorithm in SMOKE

• Upgrade to latest Models-3/CMAQ release (July 2002)–Computer time issue (runtime doubled)–Nitrate PM overprediction issue–Organic PM/Biogenic SOA issue

On-going and Upcoming Modeling On-going and Upcoming Modeling Activities at OAQPS/AQMG Activities at OAQPS/AQMG (contd.)(contd.)

((JanuaryJanuary avg.) avg.)

Difference (CMAQ/2002 – CMAQ/2001)

PM 2.5 PM 2.5

Nitrate Nitrate PMPM

((JanuaryJanuary avg.) avg.)

Nitrate PM

CMAQ-2001 (AE2)CMAQ-2001 (AE2) CMAQ-2002 (AE3) CMAQ-2002 (AE3)

((JulyJuly avg.) avg.)

Difference (CMAQ/2002 – CMAQ/2001)

PM 2.5 PM 2.5

Biogenic Biogenic SOASOA

CMAQ-2002(AE3)

CMAQ-2001(AE2)

Anthropogenic SOA Biogenic SOA

(July avg.)(July avg.)

2 ug/m3 6 ug/m3

Annual National CMAQ run for Annual National CMAQ run for year 2001year 2001• 2001 MM5 & NEI emissions thru SMOKE

• Model evaluation against expanded PM speciation network (TREND, IMPROVE, etc.)

““Intercontinental Transport of Climatic Intercontinental Transport of Climatic Effect of Air Pollutants” (Effect of Air Pollutants” (ICAPICAP) Project) Project• Modeling of trans-oceanic pollutant transport

• Nested modeling over East Asia/China Region

Upcoming Modeling Activities at Upcoming Modeling Activities at OAQPS/AQMG OAQPS/AQMG

“Intercontinental Transport and Climatic Effects of Air Pollutants

(ICAP)” : USEPA Modeling Initiative

NASA/MOPPIT: CO Transport

Intercontinental Transport and Climatic Effects of Air Pollutants

(ICAP)

Air Quality Modeling over China : Ozone

PM 2.5 Sulfate PM

Organic PMNitrate PM

Air Quality Modeling over China : PM

Visibility (Deciview) Sulfate Wet Deposition

Air Quality Modeling over China

PM 2.5 modeling more complex than O3 modeling• Multi-species vs. single species

• SO2, NOx, VOC, or NH3-limited vs. VOC or NOx-limted

• Annual vs. summer pollutants

• Day + nighttime chemistry vs. daytime chemistry

• “Gas + aqueous + aerosol” chemistry vs. gas chemistry

• Organic PM modeling is still a research issue

• Emission (e.g. NH3, fire, Biogenic) & monitoring issues (e.g., OC/EC, nitrate) and provide further modeling challenge

O3 and PM are highly linked, “one-atmosphere” modeling is necessary

Take Home MessageTake Home Message

Nitrate PM Nitrate PM Sulfate PM Sulfate PM ---- (decrease) -- (decrease)(decrease) -- (decrease)

((JulyJuly Average, 1996) Average, 1996)

50% NOx reduction

Formation of Secondary PM (PM2.5):

Sulfate PM formation: H2SO4 + 2 NH3 ---> (NH4)2SO4 (s)

Gas Phase: O2,H2O

SO2 + OH ---> H2SO4

Aqueous Phase: H2O

S(IV) + H2O2 ---> H2SO4 (Dominate over low pH)

S(IV) + O3 ---> H2SO4

Nitrate PM formation: HNO3 + NH3 <---> NH4NO3

(aq,s)

Gas Phase : (daytime)

NO2 + OH ---> HNO3

Gas &Aq Phase : (nighttime)

N2O5 + H2O ---> HNO3

Organic PM formation:

(Gas-Particle conversion)

VOC + OH ---> SOA O3, NO3 (semi-volatile)

(Long-chain VOC, Aromatics, Biogenic VOC)

Ozone Ozone Sulfate PM Sulfate PM ---- (decrease) -- (decrease)(decrease) -- (decrease)

((JulyJuly Average, 1996) Average, 1996)

50% NOx reduction

Ozone Ozone Organic PM Organic PM ---- (decrease) -- (decrease)(decrease) -- (decrease)

((JulyJuly Average, 1996) Average, 1996)

50% NOx reduction

Nitrate PM Nitrate PM Sulfate PM Sulfate PM -- (decrease) + (increase)-- (decrease) + (increase)

((JanuaryJanuary Average) Average)

50% NOx reduction

Ozone Ozone Sulfate PM Sulfate PM + (increase) + (increase)+ (increase) + (increase)

((JanuaryJanuary Average) Average)

50% NOx reduction

Air ToxicsAir Toxics

OzoneOzone

Acid RainAcid Rain

VisibilityVisibility

PMPM2.52.5

Water Water QualityQuality

..OHOH

..OHOH Role in Linking Pollutants Formation : Role in Linking Pollutants Formation : One-AtmosphereOne-Atmosphere

NOx + VOC + OH + hv ---> O3

SOx [or NOx] + NH3 + OH ---> (NH4)2SO4 [or NH4NO3]

SO2 + OH ---> H2SO4

NO2 + OH ---> HNO3

VOC + OH --->Orgainic PM

OH <---> Air Toxics(POM, PAH, Hg(II), etc.)

Fine PM(Nitrate, Sulfate, Organic PM)

NOx + SOx + OH (Lake Acidification,

Eutrophication)