Presentation is loading. Please wait.

Presentation is loading. Please wait.

Inverse modelling of CO emissions J.-F. Müller and T. Stavrakou Belgian Institute for Space Aeronomy Avenue Circulaire 3, 1180 Brussels

Similar presentations


Presentation on theme: "Inverse modelling of CO emissions J.-F. Müller and T. Stavrakou Belgian Institute for Space Aeronomy Avenue Circulaire 3, 1180 Brussels"— Presentation transcript:

1 Inverse modelling of CO emissions J.-F. Müller and T. Stavrakou Belgian Institute for Space Aeronomy Avenue Circulaire 3, 1180 Brussels jfm@aeronomie.be EVERGREEN International Workshop 19-20 January 2006, KNMI, De Bilt, The Netherlands

2 Outline Carbon monoxide: sources and sinks Inverse modeling of emissions using the adjoint model State-of-the-art in CO inversion The IMAGES model used in two inversion exercises constrained by:  a) 1997 CMDL data & GOME NO 2 columns  b) the 2000-2001 MOPITT CO columns Big-region vs. grid-based inversion approach Comparison to independent observations and past studies Conclusions and outlook

3 Carbon monoxide: sources and sinks COCO 2 CH 2 O CH 4 OHOH, hvOH 1100570360 8530 deposition NMVOC (non-methane volatile organic compounds) 700 100 50 200 80 250 OH,O 3 100 340 deposition SOA= Secondary Organic Aerosols CO 2 (units: Tg C/year) 410 ?? ?

4 Inverse modelling of emissions Cost function: measures the bias between the model and the observations J(f)=½Σ i (H i (f)-y i ) T E -1 (H i (f)-y i ) + ½ (f-f B ) T B -1 (f-f B ) Model operator acting on the control parameters observations 1st guess values of the control parameters Matrix of errors on the observations Matrix of errors on the control parameters Vector of the control parameters For what values of f is the cost function minimal?

5 The adjoint model Gradient of the cost function Calculation of new parameters f with a descent algorithm Minimum of J(f) ? Observations Forward CTM Integration from t 0 to t Transport Chemistry Cost function J(f) Adjoint model Integration from t to t 0 Adjoint transport Adjoint chemistry Adjoint cost function Current informations Control variables f yes no Optimized control parameters

6 Inversion studies Model used Observations used Bergamaschi et al., 2000TM2CMDL 1993-1995 Pétron et al., 2002IMAGESCMDL 1990-1996 Kasibhatla et al., 2002GEOS-CHEMCMDL 1994-1996 Palmer et al., 2003GEOS-CHEMTRACE-P 2001 Arellano et al., 2004GEOS-CHEMMOPITT 2000 Pétron et al., 2004MOZARTMOPITT 2000-2001 Müller & Stavrakou, 2005IMAGES + ADJOINTCMDL 1997 GOME NO2 col. 1997 Pétron et al., to be submittedMOZARTMOPITT 2000-2004 Stavrakou & Müller, submittedIMAGES + ADJOINTMOPITT 2000-2001 Inverting for CO emissions – State-of-the-art

7 Advantages from the use of the adjoint The calculated derivatives are exact The full (transport/chemistry) adjoint allows to take non-linearities into account, e.g. the non-linear relationship between CO concentrations and surface emissions The emissions of different compounds can be optimized simultaneously, their chemical interactions being taken into account The computational time to determine the model sensitivity does not depend on the number of control variables  grid-based inversions can be addressed BUT: the exact posterior error estimation is not possible within this framework Instead, iterative approximations of the inverse Hessian can be used

8 The IMAGES model Provides the global distribution of 60 chemical compounds at 5°x5° resolution and 25 vertical levels (Müller and Brasseur, 1995) A priori anthropogenic emissions : 1997 EDGAR v3 inventory (Peters and Olivier, 2003) Biomass burning emissions : GFED database (Van der Werf et al., 2003) or the POET inventory (Olivier et al., 2003) Biogenic emissions for isoprene and monoterpenes from Guenther et al., 1995, and for CO from Müller and Brasseur, 1995 Model time step : 1 day, spin-up time : 4 months, 1 year simulation

9 A. Big-region inversion of the 1997 CO emissions The inversion is constrained by: NOAA/CMDL CO mixing ratios Ground-based FTIR CO vertical column abundances GOME tropospheric NO 2 columns Simultaneous optimization of the total annual CO & NOx emissions over large regions (39 flux parameters) chemical feedbacks via the adjoint constant seasonality of the sources B is assumed diagonal Müller and Stavrakou, ACP, 2005

10

11 Impact of emission changes on OH

12 Comparison to aircraft observations

13 Direct calculation of the Hessian matrix using finite differences on the adjoint model Use of the inverse BFGS formula and the output of the minimization algorithm at each iteration Use of the DFP update formula Estimation of the posterior errors

14 B. Big-region vs. grid-based inversion for optimizing the 2000-01 CO&VOC emissions The inversion is constrained by the MOPITT daytime CO columns from May 2000 to April 2001 The columns and averaging kernels are binned onto the IMAGES grid and monthly averaged  total : ~ 6000 observations Error on the column is assumed 50% of the observed value « Big-region approach »: optimize the global CO fluxes over large regions as in case A (18 variables) « Grid-based » inversion: optimize the fluxes emitted from every model grid cell by month ( ~30000 param.) seasonality and geographical distribution varied source-specific correlations among prior errors on the flux parameters  B non-diagonal In both cases, distinguish between anthropogenic, biomass burning and biogenic emissions Stavrakou and Müller, 2006, submitted

15 The error correlation setup Anthropogenic emissions errors: highly correlated within the same country weakly correlated within large world zones uncorrelated in any other case constant temporal correlation Vegetation fire and biogenic emissions: spatial correlations decrease with geographical distance they are further reduced when the fire or ecosystem type differ temporal correlations

16 Optimization results Both solutions succeed in reducing the model/MOPITT bias over most regions Larger cost reduction in the grid- based case (4.6) as compared to the big-region setup (2.2) Big-region setup Grid-based setup MOPITT column

17 Anthropogenic emission updates  Optimized global anthropogenic emissions : 664 Tg CO/yr (+16%)  More significant increase over the eastern China in the grid-based (110%), compared to the big-region setup (80%)  Reduced South Asian emissions by ~40%  Small changes over America, Europe and Oceania Big-region setupGrid-based setup

18 Anthropogenic emissions by region

19 Vegetation fire emission updates Big-region setup Grid-based setup Seasonal variation prior GFED prior POET big-region GFED grid-based GFED grid-based POET Remarkable convergence of optimizations using either GFED or POET prior emissions Important changes in seasonality of biomass burning emissions Increased S. African emissions in September, reduction in June when using GFED

20 Biogenic emission updates Seasonal variation  Global enhancement of biogenic VOC emissions (~ +15%)  Higher NMVOCs oxidation source by 10% grid-based inversion prior big-region grid-based

21 Comparison to independent data (CMDL, FTIR, aircraft campaigns) prior big-region grid-based

22 Comparison of our results to past inverse modelling studies

23 Conclusions and perspectives Feasibility of the multi-compound inversion Higher performance of grid-based inversion for reactive compounds Importance of the error correlation setup for better constraining the large number of emission parameters in the grid-based framework The posterior uncertainty analysis (using the DFP approximation) shows important error reductions for large-scale fluxes (e.g. Chinese anthropogenic emissions, African biomass burning), but small error reductions for individual grid cells Large increases of anthropogenic emissions over Far East Synergetic use of different datasets is required to better quantify emissions, in particular the CO production from the NMVOCs


Download ppt "Inverse modelling of CO emissions J.-F. Müller and T. Stavrakou Belgian Institute for Space Aeronomy Avenue Circulaire 3, 1180 Brussels"

Similar presentations


Ads by Google