A Posteriori NO x Emissions molecules/cm 2 s 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 x 10 11 Modeled tropospheric vertical NO 2 columns molecules/cm 2 1 2 3 4 5 6 7 8 9 10 11 x 10 15 Recent advances in applying satellite observations for inverse modeling of surface NO x emissions Akhila Padmanabhan 1 , Nicolas Bousserez 1 , Randall V. Martin 1,2 , Daven K. Henze 3 , Monika Kopacz 4 , Kumaresh Singh 5 , Changsub Shim 6 , Kevin Bowman 7 , Adrian Sandu 5 Introduction •Emissions of nitrogen oxides, (NO x = NO +NO 2 ), play a major role in air quality and climate change. •Direct inference of emissions is impeded by sparse in situ measurements. •Satellites provide broad spatial and temporal perspectives that could improve our understanding of emissions through inverse modeling Satellite Observations of NO 2 NO x Emissions • Mass Balance Approach • Adjoint Approach Inverse Modeling Observations from SCIAMACHY •SCanning Imaging Absorption spectroMeter for Atmospheric CartograpHY •Aboard the Envisat Satellite •Measure direct and backscattered solar radiation •NO 2 retrieved as tropospheric columns following Lamsal et al. [2011] Mass- Balance Approach • Top -down emissions inventory is inferred by: •The top-down and bottom up emissions inventory are then combined , weighted by relative errors Advantages: • computationally less expensive • easily implemented for a variety of forward models r t E 2 2 2 2 ) (ln ) (ln ) )(ln (ln ) )(ln (ln ln t a t a a t E E E E= a posteriori emissions E t = top- down emissions E a = bottom- up emissions Ω a =model NO 2 column Ω r =observed NO 2 column α = E a / Ω a ε b = relative error in bottom up estimate ε a = relative error in top down estimate A Posteriori NO x Emissions molecules/cm 2 s 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 x 10 11 The Adjoint Approach • parameter scaling factors are applied to a priori emissions •seeks a σ value that minimizes the cost function, J: Advantages: • the optimization better exploits the dynamic (chemistry, transport) of the model • explicitly accounts for spatial smearing i i i a E E , ) ( ) ( 2 1 )... ( ) ( 2 1 1 1 a a T a r obs k c obs T obs S Hc S Hc J σ = emission scaling factors c= modeled vertical NO 2 profiles Ω obs = observed NO 2 columns H = observational operator S obs = observational error covariance matrix γ r = regularization parameter (=1) σ a = a priori estimate of emissions scaling factor S a =error covariance estimate of emissions scaling factors (assumed to be 100% globally) i= source k= domain (time, space, chemical species) Comparison •Absolute difference between a posteriori and a priori emissions in adjoint approach over Europe and United States shows high degree of variability between adjacent pixels which is not observed in the mass balance approach • Values for NO x emissions generally higher when derived from mass balance approach for areas where anthropogenic emissions do not represent a dominant source •Need to compare two approaches for longer period of time in order to better address these differences • Future work will also include a seasonal comparison of the methods, using a higher resolution, and using observations from other satellites, such as OMI References Henze, D. K., A. Hakami and J. H. Seinfeld (2007), Development of the adjoint of GEOS-Chem, Atmos. Chem. Phys., 7, 2413-2433. Henze, D. K., J. H. Seinfeld and D. T. Shindell (2009), Inverse modeling and mapping U.S. air quality influences of inorganic PM2.5 precursor emissions with the adjoint of GEOS-Chem, Atmos. Chem. Phys., 9, 5877-5903. Lamsal, L. N., R. V. Martin, A. Padmanabhan, A. van Donkelaar, Q. Zhang, C. E. Sioris, K. Chance, T. P. Kurosu and M. J. Newchurch, Application of satellite observations for timely updates to global anthropogenic NOx emission inventories, Geophys. Res. Lett., 38, L05810, doi:10.1029/2010GL046476, 2011 Martin, R.V., D.J. Jacob, K. Chance, T.P. Kurosu, P.I. Palmer, and M.J. Evans, Global inventory of nitrogen oxide emissions constrained by space-based observations of NO 2 columns, J. Geophys. Res., 108(D17), 4537, doi:10.1029/2003JD003453, 2003. 1 Dalhousie University, Halifax, Nova Scotia, Canada 2 Harvard-Smithsonian Center for Astrophysics, Cambridge, Massachusetts, USA 3 Department of Mechanical Engineering, University of Colorado, Boulder, CO, USA 4 Woodrow Wilson School of Public and International Affairs, Princeton University, Princeton, New Jersey, USA 5 Department of Computer Science, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, USA 6 Korea Environment Institute, Republic Of Korea 7 Jet Propulsion Laboratory, Pasadena, CA, USA Supported by NSERC Relative Error % 0 50 100 150 Tropospheric NO 2 vertical columns derived from SCIAMACHY for November 15-21, 2006 (left) , and the relative errors (above) used in both approaches A Priori NO x Emissions molecules/cm 2 s 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 x 10 11 Forward Model (GEOS-Chem) • a priori NO x emissions from anthropogenic, biomass burning, biofuel, and soil sources produced for same time period as observations, error assumed to be 100% uniform error • solves for NO 2 columns resulting from a priori emissions, assumed to be 30% uniform error • both emissions and columns at 4° X 5° resolution, GEOS5 met fields A Posteriori/A Priori NO x Emissions unitless 0 0.5 1 1.5 2 2.5 3 A Posteriori- A Priori NO x Emissions molecules/cm 2 s -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x 10 11 A Posteriori/A Priori unitless 0 0.5 1 1.5 2 2.5 3 A Posteriori-A Priori molecules/cm 2 s -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x 10 11 A Posteriroi (Adjoint)/A Posteriori (Mass Balance) unitless 0 0.5 1 1.5 2 2.5 3 A Posteriori(Adjoint)-A Posteriori(Mass Balance) molecules/cm 2 s -4 -3 -2 -1 0 1 2 3 4 x 10 11 2 molecules/cm 2 0 1 2 3 4 5 6 7 8 x 10 15