Top Banner
High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia Institute of Technology Oct 11, 2010
15

High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Mar 29, 2015

Download

Documents

Dandre Setters
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ

Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell

Georgia Institute of TechnologyOct 11, 2010

Page 2: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

PM has the potential to affect

- Visibility- Human health- Climate…

Page 3: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Sensitivity Analysis

• A tool to quantify the response of pollutant concentrations to the changes in source emissions or other parameters

• Sensitivity can be calculated by different methods, depending on the focus of application.

- Brute Force Method (BFM) - Decoupled Direct Method (DDM) - Adjoint

Air QualityModel

Input:E, T, RH, U, R…

Output:Ci

Sensitivity

What’s the response?

Page 4: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Decoupled Direct Method

Air QualityModel

E C(E)Base Run

Air QualityModel

E - ΔE C(E - ΔE)Third Run

Air QualityModel

E + ΔE C(E + ΔE)Second Run

(1) ( ) ( )

2BFM

C E E C E ES

E

First Order

(2)2

( ) 2 ( ) ( )

( )BFM

C E E C E C E ES

E

Second Order

Air Quality Model

C aerosol C transportC gas

S aerosol S transportS gas

S0

C

S

C0

First Order

(1),

( )

( )i

i pp

ConcentraitonS

Emission

1 2

2(2), ,

1 2

( )ii p p

ConcentrationS

Ep Ep

Second Order

BFM DDM-PM

Page 5: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Brief History of DDM Sensitivity in Air Quality Models

• 1980’s: Dunker developed for air quality modeling

- Gas phase, first order• 1997+: Implementation in CIT, URM, CMAQ, CAMx• 2002+: First order aerosol phase (eg., Boylan et al., 2002 in

URM; Napelenok et al., 2006 in CMAQ; Koo et al., 2007 in CAMx)

• 2003: High order gas phase (eg., Hakami et al. 2003 in CMAQ)

- Use for source apportionment and uncertainty analysis

Page 6: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Implementing HO DDM-PM in CMAQ:

( ) ( )ii i i i

CuC K C R E

t

(2)(2) (2) (2) (1) (1)( ) ( ) '( , , )ijkijk ijk ijk i ij ik

SuS K S JS f C S S

t

(1)(1) (1) (1)( ) ( ) ( )ijij ij ij i

SuS K S JS f C

t

(Solve using the same algorithm for concentrations)

First Order

Second Order

Page 7: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

ISORROPIA

H2SO4, NH3, HNO3, HCl, H2Ov

H+, NH4+, Na+, Cl-, SO4

=, HSO4-,

NO3-, OH-, H2O,

NH3(aq) , HNO3(aq) , HCl(aq)

Salts

,( ) ( )i j

ii

a K j

RH, temperatureTotal concentrations

Sulfate-ammonium-nitrate-chloride-sodium system

i i ia x

i ia p

Equilibrium equations

Page 8: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Calculating Second-Order (and higher)DDM Sensitivities

3 3 3 2

3 3 2

2 2 2 223 3 3 2

1 2 3 1 2 3 1 2 1 2 2 1 2

21

[ ] [ ] log [ ]1 [ ] 1 1 22ln10 0

[ ] [ ] [ ] [ ]

[ ] [ ] log [ ]1 [ ] 1 1 22ln10

[ ] [ ] [ ] [ ]

1 [ ] [

[ ]

HNO

HNO

NO HNO H OH

H p NO p HNO p p H O p

NO HNO H OH

H p p NO p p HNO p p p p H O p p

H

H p

3 3 3 3 2 2

2 2 22 3 1 2 3 1 2 2 1 2

[ ] [ ] [ ] [ ] [ ] [ ]] 1 1 2

[ ] [ ] [ ]

NO NO HNO HNO H O H OH

p NO p p HNO p p H O p p

3

23

4 23 2

[ ][ ]

[ ][ ]HNOH NO

KHNO H O RT

Equilibrium Equation

First-Order Sensitivity Equation

Second-Order Sensitivity Equation

Second-Order Sensitivities;Higher order found from similar

Processes (eqs. very similar to 1st order)

263

1

log_ ( ) 1( )HNO

i

coef i S ip

2621

[ ]_ ( ) 1( )

i

H Ocoef w i S i

p

2 263

11 2

log_ ( ) 2( )HNO

i

coef i S i RHSp p

2

2621

1 2

[ ]_ ( ) 2( )

i

H Ocoef w i S i

p p

Available from 1st order

Page 9: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Evaluation of DDM Performance: First Order • RH = 95% T = 298K• Total Sulfate: 0.1~10μmol/m3

• Total Ammonium: 0.1~10μmol/m3

• Total Nitrate: 0.1~10μmol/m3

• Total Sodium: 0.5μmol/m3

• Total Chloride: 1.0μmol/m3

[ ]A

p

H+ NH4+ SO4

= HSO4- NO3

- Ap

TS

TA

TN

Brute Force Sensitivities (mol/mol)

DD

M S

ensi

tiviti

es (m

ol/m

ol)

Page 10: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Evaluation of DDM Performance: Second Order • RH = 95% T = 298K• Total Sulfate: 0.1~10μmol/m3

• Total Ammonium: 0.1~10μmol/m3

• Total Nitrate: 0.1~10μmol/m3

• Total Sodium: 0.5μmol/m3

• Total Chloride: 1.0μmol/m3

2

2

[ ]A

p

H+ NH4+ SO4

= HSO4- NO3

- Ap

TS

TA

TN

DD

M S

ensi

tiviti

es (m

ol/m

3)-1

Brute Force Sensitivities (mol/m3)-1

Page 11: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Accuracy: DDM versus BFM

• Noisy BFM second-order sensitivities

- Perturbation (Δ) - Convergence criteria (ε)

10

0.01

1 10

10

0.001

1 10

3

0.01

1 10

Page 12: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

First Order Sensitivity of AS to SOx Second Order Sensitivity of AS to SOx

Performance of HO DDM-PM in CMAQ Aerosol Sulfate (AS)

- Episode: Jan1 – Jan 7, 2004- Grid resolution: 36km, 13 vertical layers- Meteorology: MM5- Emissions: SMOKE

Page 13: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Evaluation of HO DDM-PM in CMAQ Aerosol Sulfate (AS)

First Order Sensitivity of AS to SOx Second Order Sensitivity of AS to SOx

Date: Jan 2, 2004Emission change for BFM: 50%

Page 14: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

Summary

• High-order DDM-PM has been developed and implemented in CMAQ.

• HO DDM-PM appears to perform better than BFM. - Need better way to test as BFM is subject to noise

• Results from CMAQ captures the features of physical and chemical processes.

• Future Applications - Source Apportionment - Emission inventory assessment - Air quality model uncertainty analysis

Page 15: High-Order DDM Sensitivity Analysis of Particular Matter in CMAQ Wenxian Zhang, Shannon Capps, Yongtao Hu, Athanasios Nenes, and Armistead Russell Georgia.

• Advisor: Dr. Armistead Russell• Funding Source: Conoco-Phillips• Support from CAMx modeling group

Acknowledgements