220
X. Zhang et al.
tropospheric chemical state that is better constrained. Our experiments also evaluate
the inferred CO emission estimates from major anthropogenic, biomass burning and
biogenic sources.
34.1 Introduction
Surface carbon monoxide (CO) reacts through CO–hydrocarbon–NO x chemistry
cycle, affecting tropospheric ozone (O 3 ) and hydroxyl radical (OH) abundances.
The CO emission estimates using various chemical transport models (CTMs) largely
depend on the uncertainties of the employed bottom-up inventories as well as CTMs’
uncertainties on chemistry and transport schemes [2]. In fact, many of the differences
on regional emission estimates are due to model’s chemistry scheme, especially the
modelled OH and its precursor O 3 . The modelled OH, serving as the cleansing agent
of the troposphere, would determine the atmospheric concentrations of CO, NonMethane Volatile Organic Compound (NMVOC), methane etc., and affect estimated
CO emissions. To reduce the model’s uncertainties due to chemistry, considering the
significant chemical interactions of the tropospheric CO–hydrocarbon–NO x –O 3 –OH
cycle, it is essential to establish strong chemical constraints on multiple tropospheric
constituents. Multiple Species data Assimilation (MSA) has shown potential to better
estimates CO and NO x emissions via improving OH and its precursor O 3 . Since building a chemical consistent framework throughout the assimilation would be conducive
to understand both model sensitivity to certain chemicals and model uncertainties. In
this paper, we use the GEOS-Chem 4D-Var data assimilation system to examine the
impact of assimilating multiple chemical species to quantify CO surface emissions.
We assimilated observations of O 3 from IASI and OSIRIS, NO 2 from OMI, and CO
from MOPITT with two-week assimilation window running in 2016.
34.2 Methods
In our experiment, full chemistry version of GEOS-Chem adjoint model with 4Dvar data assimilation scheme is used. The details of the inversion methodology are
explained by Henze et al. [1]. Full chemistry version of the model together with 4Dvar assimilation scheme would enable chemical feedbacks among different species
throughout the assimilation window. This would allow each species to be also assimilated by observations of other related tracers. The full list of emission inventories
followed v09 of the GEOS-Chem model. Our study uses GEOS-FP meteorological
fields, with 4 × 5° horizontal resolution. The spin-up period of all the chemical tracers
is one month starting January-2016. All instrument run mentioned in the improved
MSA optimizes CO emission, isoprene emissions, surface NO x emissions as well as
O 3 and HNO 3 initial conditions when all observations mentioned in Table 34.1 were
X. Zhang et al.
tropospheric chemical state that is better constrained. Our experiments also evaluate
the inferred CO emission estimates from major anthropogenic, biomass burning and
biogenic sources.
34.1 Introduction
Surface carbon monoxide (CO) reacts through CO–hydrocarbon–NO x chemistry
cycle, affecting tropospheric ozone (O 3 ) and hydroxyl radical (OH) abundances.
The CO emission estimates using various chemical transport models (CTMs) largely
depend on the uncertainties of the employed bottom-up inventories as well as CTMs’
uncertainties on chemistry and transport schemes [2]. In fact, many of the differences
on regional emission estimates are due to model’s chemistry scheme, especially the
modelled OH and its precursor O 3 . The modelled OH, serving as the cleansing agent
of the troposphere, would determine the atmospheric concentrations of CO, NonMethane Volatile Organic Compound (NMVOC), methane etc., and affect estimated
CO emissions. To reduce the model’s uncertainties due to chemistry, considering the
significant chemical interactions of the tropospheric CO–hydrocarbon–NO x –O 3 –OH
cycle, it is essential to establish strong chemical constraints on multiple tropospheric
constituents. Multiple Species data Assimilation (MSA) has shown potential to better
estimates CO and NO x emissions via improving OH and its precursor O 3 . Since building a chemical consistent framework throughout the assimilation would be conducive
to understand both model sensitivity to certain chemicals and model uncertainties. In
this paper, we use the GEOS-Chem 4D-Var data assimilation system to examine the
impact of assimilating multiple chemical species to quantify CO surface emissions.
We assimilated observations of O 3 from IASI and OSIRIS, NO 2 from OMI, and CO
from MOPITT with two-week assimilation window running in 2016.
34.2 Methods
In our experiment, full chemistry version of GEOS-Chem adjoint model with 4Dvar data assimilation scheme is used. The details of the inversion methodology are
explained by Henze et al. [1]. Full chemistry version of the model together with 4Dvar assimilation scheme would enable chemical feedbacks among different species
throughout the assimilation window. This would allow each species to be also assimilated by observations of other related tracers. The full list of emission inventories
followed v09 of the GEOS-Chem model. Our study uses GEOS-FP meteorological
fields, with 4 × 5° horizontal resolution. The spin-up period of all the chemical tracers
is one month starting January-2016. All instrument run mentioned in the improved
MSA optimizes CO emission, isoprene emissions, surface NO x emissions as well as
O 3 and HNO 3 initial conditions when all observations mentioned in Table 34.1 were