Evaluation of CMAQ NO2 Predictions Over the U.S. Using Ground-based and Satellite Observations

Havala O. T. Pye1, Sergey L. Napelenok1, Robert W. Pinder1, Daven K. Henze2, Randall V. Martin3, and K. Wyat Appel11Atmospheric Modeling and Analysis Division, U.S. Environmental Protection Agency; 2Department of Mechanical Engineering, University of Colorado at Boulder; 3Department of Physics and Atmospheric Science, Dalhousie University

corresponding author: pye.havala@epa.gov

Introduction

Comparison to Surface Observations

Comparison to Satellite: Vertical Column Method Comparison to Satellite: Slant Column Method

Modeled - Observed NO212:00 to 14:59 LST

NOx Concentration by Hour AQS NO2 Concentration by Hour

CMAQ predictions are compared to hourly observations of NOx and/or NO2 from the Air Quality System (AQS) network and the Southeast Aerosol Research and Characterization Study Ex-periment (SEARCH). Model to observation comparisons reveal that:- AQS NOx tends to be fairly well predicted in the morning hours- AQS NO2 is overpredicted in the morning and evening while underpredicted in the middle of the day (note that measurements in the middle of the day are likely biased high due to interference from other reactive nitrogen)- At SEARCH sites (N=7), which should have little measurement error, CMAQ tends to slightly overpredict the NO2 concentrations in the early afternoon

OMI flyover time: early afternoon (~1-2pm)

MB: -0.9 ppbNMB: -14%

Approach:- Compare on the basis of the model’s (CMAQ’s) vertical profile- Adjust the observed satellite slant column to the model vertical column using a model-based air mass factor [Lamsal et al., 2010]

Approach:- Compare on the basis of the a priori profile- Adjust the modeled vertical column to the satellite observed slant column

Modeled vertical column: Observed vertical column: Modeled slant column: Observed slant column:

Cost Function: Adjoint forcing: Cost Function: Adjoint forcing:

Leads to more rigorous vertical column comparison. MB: -0.22x1015 molecules cm-2, NMB: -19%Emphasizes altitudes where OMI is more sensitive

MB: -0.16x1015 molecules cm-2, NMB: -15%

Emissions of NOx (NO + NO2) and HONO contribute to the formation of multiple criteria air pollutants for which the EPA sets National Ambient Air Quality Standards (NAAQS). NO2 itself has a NAAQS which was strengthened in early 2010. Through VOC-NOx chemistry, NOx emissions contribute to the formation of ozone and PM2.5, each with their own NAAQS. Current air quality models have difficulty predicting the sensitivity of ozone to NOx emission changes, and if standards for species like ozone and PM2.5 become more stringent, more accurate emission estimates and models will be needed. Comparisons of modeled and observed NO2 concentrations provide an opportunity to evaluate model performance. For applications such as data assimilation, comparisons should minimize the influence of the a priori profile used to compute satellite observed columns.

Model: - Community Multiscale Air Quality (CMAQ) v4.7.1 with lightning NOx - Continental U.S. domain at 12 km by 12 km horizontal resolution - 11 to 25 July 2006

Observations: - AQS and SEARCH ground-based NO2 and NOx - Derivation of OMI tropospheric NO2 (DOMINO) Slant Columns from the Ozone Monitoring Instrument (OMI) aboard Aura [Boersma et al., 2007]

References

Boersma, K.F., H.J. Eskes, J.P. Veefkind, E.J. Brinksma, R.J. van der A, M. Sneep, G.H.J. van den Oord, P.F. Levelt, P. Stammes, J.F. Gleason and E.J. Bucsela, Near-real time retrieval of tropospheric NO2 from OMI, Atm. Chem. Phys., 2013-2128, 2007.Lamsal, L.N., R.V. Martin, A. van Donkelaar, E.A. Celarier, E.J. Bucsela, K.F. Boersma, R. Dirksen, C. Luo, Y. Wang, Indirect validation of tropospheric nitrogen dioxide retrieved from the OMI satellite instrument: Insight into the seasonal variation of nitrogen oxides at northern midlatitudes, J. Geophys. Res., 115, D05302, 2010.

AcknowledgementsThe authors would like to thank Jenise Swall and Kristen Foley for their useful discussions regarding R. The authors would also like to thank Lok Lamsal and the CMAQ adjoint working group for discussions and coding guidance and Jesse Bash for his helpful comments. We acknowledge the free use of tropospheric NO2 column data from the OMI sensor from www.temis.nl. UC Boulder work is supported by NASA Applied Sciences (NNX09AN77G).

Summary and Conclusions

Ground-based observations indicate that CMAQ may underpredict or overpredict NO2 depending upon the hour of the day, observation network, and degree of interference from other reactive nitrogen species. Future work should examine measurement errors.

AQS ground-based and satellite observations indicate that CMAQ generally underestimates NO2 concentrations in the early afternoon but overpredicts NO2 in locations such as eastern Michigan, southern California, and southern Louisiana.

The two methods presented for comparing satellite observations to CMAQ predictions give the same direction of disagreement (underestimate vs overestimate), but emphasize different quantities.

1015 molecules cm −2

11−25 July 2006

max: 45

0

5

1011−25 July 2006

max: 66

11−25 July 2006

max: 30

1015 molecules cm−2

0

5

1011−25 July 2006

max: 32

10-4 10-2 100 102 104 10-15 molecules-1cm2

−10

−5

0

5

10

min: −29

max: 13

60

-60

-40

-20

0

20

40

OMI tropospheric slant NO2 column

1015 molecules cm−2

17 July 2006

max: 320

5

10

15

20

Satellite Air Mass Factors

Air mass factor that depends on CMAQ:DOMINO provided air mass factor: OMI tropospheric air mass factora priori NO2 profileCMAQ subcolumn of NO2 on level Averaging kernel for level Estimated error, set to (1x1015 molec/cm2)2

Cost functionOMI troospheric slant columnVertical column

11−25 July 2006

0.5

1.0

1.5

2.0

2.511−25 J uly 2006

0.5

1.0

1.5

2.0

2.5

The total air mass factor depends on an a priori NO2 vertical profile and can be used to convert slant columns to vertical colums (Eqn 1). An averaging kernel provides a link between a retrieved value and an actual vertical column value. To remove the dependence on the a priori profile in model-observation comparisons, a CMAQ based air mass factor can be computed:

OMI offers:- potentially complete spatial coverage on a daily basis- measured NO2 columns uncontaminated by other reactive nitrogen species

Modeled-Observed 10-4 10-2 100 102 104

[10-15 molecules-1cm2]

[10-15 molecules-1cm2]

~ (Modeled-Observed)2

Relationship between vertical and slant column (Eqn 1):

[ppb]

SEARCH NO2 Concentration by Hour

MB: 3.1 ppbNMB: 23%

MB: 3.2 ppbNMB: 32%

MB: 1.1 ppbNMB: 15%

(triangle)

(circle)

Mean Bias (MB) between Modeled (Mi) and Observed (Oi) values:

Normalized mean bias (NMB):