Belgian Institute for Space Aeronomy

Avenue Circulaire 3, B-1180 Brussels

BenefitsBenefits fromfrom the the combinedcombined use of satellite use of satellite observations and inverse modelling : observations and inverse modelling : achievementsachievements, ,

limitations, and perspectives limitations, and perspectives

Modelling Unit : Jenny Stavrakou and J.-F. MullerThanks to our data providers :BIRA/IASB : I. De Smedt, C. Lerot, M. Van RoozendaelInstitute for Remote Sensing, Univ. of Bremen : M. Vrekoussis, F. Wittrock, A. Richter, J. BurrowsKNMI : F. Boersma, R. van der AThe MOPITT team

ACCENT-AT2 Follow-up Meeting, 22 - 23 June 2009

MPI Mainz

“Top-down” or inverse modelling approach

Aim : adjust the emissions used to drive a CTM so as to minimize the discrepancy between the model predictions and a set of atmospheric observations, through an inversion method Wealth of satellite products provided :

at a high viewing scene resolution & with a global coverage

Inverse modelling concrete remote sensing application : state-of-knowledge on physical&chemical atmospheric processes + advanced technical tools

used for non-reactive and reactive gases

The adjoint model : powerful tool to invert for emissions of reactive gases

address non-linearities (e.g. deriving NOx emissions from NO2 columns)

handle any large number of control parameters inversion at the model resolution

distinguish between emission categories based on prior spatiotemporal distributions

Inversion of tropospheric NO2 columns

First “top-down” global NOx emission inventory using 1 year GOME/Harvard data (Martin et al. 2003)

Lightning NOx from GOME/KNMI (Boersma et al. 2005)

First simultaneous inversion of NO2 columns from GOME/Bremen and CO/CMDL for 1997 (Muller and Stavrakou, 2005), global scale

Over N. America and China using GOME/Harvard (Martin et al., 2006, Wang et al. 2007)

Trends in the retrieved NO2 columns (Richter et al., 2005, van der A et al., 2008)

Trends inferred from inverse modelling : over Europe (Konovalov et al. 2008), on the global scale (Stavrakou et al. 2008)

Monthly mean tropospheric NO2 columns : model vs. observations

GOME SCIAMACHY

Prior Optimized Stavrakou et al., GRL, 2008Stavrakou et al., GRL, 2008

Prior & optimized NOx emissions

Strong anthropogenic emission trend over China (9%/yr) and the US, -4.3%/yr in Indiana/Ohio, less over Europe

Important trends in surface emissions trends in surface ozone :ca. +15%/decade in China in summertime

Stavrakou Stavrakou et al., GRL, et al., GRL, 20082008

GOME SCIAMACHY v1.04

Emissions over the Far East increase by 70% over the decade ~1/3 of the total anthropogenic source in 2006

N. American and European emissions reduce by ~30% and ~10%, resp.

vs.

OMI higher viewing scene resolution than SCIAMACHY (x6), global coverage in 1 day instead of 6 with SCIAMACHY, SCIAMACHY : 10:00 LT, OMI : 13:30 LT

Examine the consistency between the two instruments, study the effect of the different observing times

Compare in terms of diurnal variation of NOx emissions and chemistry

SCIAMACHY v1.1 OMI v1.0.2x1015 molecules.cm-2

Higher over urban centers morning commute, longer NOx lifetime (chemical loss peaks around noon)

Higher over regions dominated by natural sources and fires fire emissions peak in the afternoon, soil and lighning fluxes have a diurnal cycle

August 2006

SCIAMACHY v11 / OMI v1.0.2 IMAGESv2 10:00/13:30 trop. NO2 column

August 2006

1.55 (1.29)1.39N.China

1.33 (1.02)0.85S.Africa

1.49 (1.30)1.13Europe

1.20 (1.29)0.81SE US

1.32 (1.31)1.15NE US

Modelled 10:00-13:30 ratio

Observed ratio

Source regions

ratios calculated with GEOS_Chem (Boersma et al. 2008)

Two inversions performed in 2006

Inferred global emission updates by category

0

5

10

15

20

25

30

35

Anthrop. BBurning Soils Lightning

Prior SCIAv1.1 OMIv1.0.2

4% lower with SCIAMACHY, 23% higher with OMI

35% higher with OMI

close to the prior with SCIA, > x2 with OMIv1.0.2 !

close to prior

Tg N

0

2

4

6

8

10

N.America Far East Europe

Prior SCIAv1.1 OMIv1.0.2

AnthropogenicFar East : derived emissions from

SCIAMACHY 6% lower than the prior, OMI-derived emissions by 20% higher

N.America : 40% higher with OMI, contradicting previous work using SCIA(Kim et al., 2006, etc)

OMI too high compared to aircraft campaigns over N.America (Bucsela et al. 2008)

OMI agrees with gr.-based meas. from Israeli sites (Boersma et al., 2009)

Tg N

OMI-derived estimate for global soil NOx emissions (20 Tg N/yr) : much higher than the upper limit of current estimates (12 Tg N/yr, eg. Ganzeveld et al., 2002)

underestimation of the diurnal model emission profiles? further investigation

0

1

2

3

4

5

N.Africa S.Africa S.America

Soils

Proceed. AQ Conf., 2009a

NMVOCs : large variety of short-lived compounds, ozone and SOA precursors

NMVOC global burdens difficult to derive, serious discrepancies betweeninventories

large uncertainties in the speciation & global modelling of highly reactive gases

New possibilities from space measurements : constrain NMVOC emissions using HCHO columns

Potential of spaceborne HCHO columns to provide quantitative information about NMVOC emissions (Palmer et al., 2003, 2006, Fu et al., 2007, Dufour et al., 2009)

De Smedt et al., ACP, 2008, www.temis.nl

Summer 2006

Stavrakou et al., ACP, 2009b

SCIAMACHY HCHO IMAGESv2 HCHO

Perform inversion of SCIAMACHY HCHO columns on the global scale(Stavrakou et al., ACP, 2009c)

Optimized/prior isoprene emission ratio

large biogenic flux increase (55%)

30% decrease of biogenic fluxes35% reduction,

supported by OMI-derived fluxes (Millet et al., 2008)

MEGAN-ECMWF (Muller et al., 2008) Sensitivity of isoprene estimatesto the choice of inversion setup : very

weak

to the use of OH recycling in theisoprene mechanism as suggested by Lelieveld et al., 2008 : low

to the use of the GEOS-Chem isoprenemechanism : strong, global isoprenesource higher by 38%, even higher in lowNOx regions in the Tropics, due to loweryield from isoprene in GEOS-Chem

currently incomplete knowledge about the isoprene degradation scheme

Current limitation : lack of aerosolcorrection in the retrieval might lead to overestimated columns over fire scenes

A step forward : perform multi-compound inversions

using datasets from one or several satellite instruments

one-year or multi-year

Benefits:take advantage of the constraints offered by different datasets and

assess their consistency

simultaneously optimize the emissions of different chemical species while accounting for chemical feedbacks between them

as different species have common sources, information obtained from measurements of one compound can help constrain the sources of other species

Same sounder : HCHO and CHOCHO from

2005 CHOCHO vertical columns (1014 molec. cm-2) annually averaged

Wittrock et al., GRL, 2006

Novel approach: use inverse modelling to quantify the missing terrestrial source of glyoxal needed to explain the mismatch between the observations and the model

introduce an unknown UVOC species of biogenic origin distributed as the LAI

UVOC is either direct glyoxal flux or an unknown glyoxal precursor

Derivation of the updated global glyoxal source over land

Primary UVOC source : 36 Tg/yr Secondary UVOC source : 54 Tg/yr

The additional glyoxal source

1010 molecules cm-2 sec-1

The extra amount of CHOCHO needed from terrestrial sources is equally largeas currently known terrestrial sources

Based on a better match of the model to in-situ CHOCHO observations in the remote terrestrial BL CHOCHO must be released during transport and non-colocated to spatial patterns of primary sources

Satellite data provide important constraints in order to test and improve therepresentation of SOA formation in large-scale models

Stavrakou et al., ACPD, 2009d

Different sounders : CO from and HCHO from

Multi-year CO inversion study (2003-2008)CO HCHO

CO inversion updates improve model/HCHO data agreement in the Tropics consistency

Potential to perform simultaneous inversion of CO&HCHO data

2006 Optimized/prior annually averaged emission updates

Optimized/prior biom. burning emission ratio

Inversion using COOptimized/prior biom. burning emission ratio

Inversion using HCHO

Optimized/prior biogenic emission ratio Optimized/prior biogenic emission ratio

from Stavrakou et al. ACP, 2009c

Long-time series of satellite products offer an unprecedented opportunity for continuous monitoring ofthe state of the atmosphere, potential trends

Measurements at different overpass times : test diurnal cycle of emissions

Discrepancies between the datasets are reflected in differences in the derived emissions, e.g. the enhanced OMI-derived NOx emissions lead to an unreasonably low tropospheric methane lifetime (7.4 yrs)

need for consistent data series

Misrepresentation/absence of processes in the model keep the models updated with current scientific findings

Space-based HCHO crucial information on the pyrogenic NMVOCs & isoprene, BUT not for anthrop. NMVOCs due to low signal/noise ratio

Satellite maps suggest the existence of a glyoxal terrestrial source much higher than derived based on the current knowledge of the processes use inverse modelling to quantify this extra source, enhanced oceanic CHOCHO new in-situ measurements

Simultaneously invert for emissions of different compounds is appealing but challenging topic, promising preliminary results from CO & HCHO

IASI and GOME-2 data offer a high research potential for future studies, new products will be made available soon

Identify the discrepancies between satellite products from different retrieval groups throughintercomparisons & validation with ground-based networks

Concluding remarks, open issues & perspectives

Thank you very much for yourattention !

Thank you very much for yourattention !

www.aeronomie.be/tropo

jenny@aeronomie.be

Diurnal cycles of emissions

Jenkin et al., 2000 based on Giglio, 2007

land

ocean

based on soil temperatureYienger and Levy, 1995

Sensitivity of isoprene updates to the inversion setup / isoprene scheme

Lelieveld et al., Nature, 2008

GEOS-Chemisoprene scheme(Evans et al. 2003)

Inversion setup

Standard inversion

Isoprenescheme

weak sensitivity to the choice of the inversion setup

OH recycling: lowsensitivity

GEOS-Chem: global isoprene source higherby 38%, even higher in low NOx regions in the Tropics, due to loweryield from isoprene in GEOS-Chem

strong sensitivity to the choice of theisoprene mechanismcurrently incompleteknowledge

under investigation in collaboration with KULCurrent limitation : lack of aerosol correction in the retrieval

lower retrieved columns?