1. Retrieval of CHOCHO from MAX-DOAS measurements in the Beijing area
F. Hendrick1, C. Gielen1, C. Lerot1, T. Stavrakou1, I. De Smedt1, C. Fayt1, C. Hermans1, J.-F. Müller1, G. Pinardi1, R. Volkamer2, P. Wang3, and M. Van Roozendael1
1Belgian Institute for Space Aeronomy (IASB-BIRA), Brussels, Belgium
2Dept. of Chemistry and Biochemistry, University of Boulder, Colorado
3Institute of Atmospheric Physics/CAS, Beijing, China
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

2. Outline Introduction – glyoxal in the atmosphere
CHOCHO vertical profile retrieval from MAXDOAS observations in the Beijing area
Description of correlative data sets:
CHOCHO VCD from OMI and from the 3D-CTM IMAGES
Results:
CHOCHO seasonal and diurnal variations
Glyoxal : Formaldehyde ratio
Optimal Estimation versus geometrical approximation
Conclusions
This is the outline of my presentation. In a short introduction, I will briefly discuss the importance of glyoxal
Then I will come to the results and finally I will give some conclusions
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

3. Introduction: Glyoxal in the atmosphere
Sinks
Sources
Photolysis
OH oxidation
Conversion to SOA
Dry and wet deposition
Oxidation of NMVOCs
CHOCHO
(lifetime: 2-4h)
Direct emissions from
fuel combustion
biomass burning
Anthr.(i), 26%
Fires (d), 12%
Fires (i), 8%
Anthr.(d), 3%
Biog.
51%
Measuring this species is of major importance in air quality issues
Glyoxal (CHOCHO) is one of the most important carbonyl compounds in the atmosphere. It is produced mainly by the oxidation of biogenic and anthropogenic non-methane volatile organic compounds (NMVOCs) which participate to the formation of tropospheric ozone and secondary organic aerosols. CHOCHO is also directly released by biomass burning and fossil fuel combustion. Measuring this species is therefore of major importance for air quality monitoring, especially given the scarcity of available CHOCHO observational data sets.
VOCs play an important role in air quality issues (e.g. production of tropospheric ozone in polluted environments)
With a small lifetime, glyoxal is an indicator of short-lived NMVOCs  Additional constraints for the quantification of their emissions.
Combined with measurements of other short-lived VOCs, it offers the potential to better distinguish anthropogenic, biogenic and pyrogenic NMVOC emissions (e.g. CHOCHO:HCHO ratio).
Glyoxal influences the atmospheric radiative budget as it is recognized as a significant source of secondary organic aerosols (SOA)  Constraints for the quantification of SOA production.
Importance to have global and high spatial resolution observations to identify local emissions and their spatial structures.
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

4. MAX-DOAS observation sites
Institute of Atmospheric Physics (06/ /2009)
(03/2010 till now )
Here we report on four years of MAXDOAS observations of HONO in the Beijing area
The period covered by the observation is
MAX-DOAS is a passive DOAS technique based on measurements of scattered sunlight at zenith and at different elevation angles towards the horizon (the so-called off-axis geometry), increasing therefore the sensitivity to absorbers present close to the ground such as HONO or NO2
Elevation angles: 2°, 4°, 6°, 8°, 8°, 10°, 12°, 15°, 30°, 90° (zenith)
Measurements from ~85°SZA sunrise to 85°SZA sunset
15’ per scan
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

5. DOAS settings Fitting window: 434-460 nm Fitted species:
CHOCHO (296 K)
NO2 (220 and 298 K)
O3 (223 K)
O4 (293 K)
H2O (296 K)
Ring effect
CHOCHO XS: Volkamer et al. (2005) at 296 K
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

6. bePRO profiling tool Clémer et al., AMT, 2010 DOAS DSCD
MAXDOAS
measurements
LIDORT
Apriori
Atm: P,T
Surface albedo
Trace gas: , profile
Aerosol: optical prop.
DOAS
CHOCHO DSCD
O4 DSCD
DSCD
Weighting functions
Optimal estimation
Aerosol vertical extinction profiles
Our MAX-DOAS retrieval algorithm is depicted in the following flow chart
Proving that we are effectively detecting HONO.
The DOAS analysis of measured absorption spectra provides HONO and O4 DSCDS
The DSCDs is the trace gas concentration integrated along the effective light path.
This first step is required because the light path length through the atmosphere depends on the aerosols
Wavelength:
O4: 477 nm
CHOCHO: 455 nm
Tropospheric HONO vertical profiles
A priori profile:
with SH=0.5km and SH=1.0 km
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

7. CHOCHO vertical profile retrieval
Xianghe, 23 June, 2010, 16:22 LT/54°SZA
Mean daytime CHOCHO profile
Xianghe ( )
Selection of ‘good’ scans based on:
RMS (SH=0.5 km gives the lowest RMS)
DOFS (>1)
Multiple scattering (O4 DSCDs) flag (Gielen et al., 2014)
we opt to only study the
temporal variation of the measured O4 DSCDs. Strong temporal
variability due to enhanced multiple scattering commonly
occurs in optically thick clouds, and is seen less for
high aerosol optical depth
The scaling height is the altitude (distance) where the vertical column decreases by a factor of e.
We use an exponentially decreasing a priori profile with a scaling height of 0.5 km
Uncertainty on HONO VCD and surface VMR:
20-25%
Uncertainty on NO2 VCD and surface VMR:
10-15%
represents the difference between the retrieved profile and the true profile due to vertical smoothing by the retrieval algorithm
The retrieval of both columns and near-surface concentrations is the main strength of the MAX-DOAS technique: it helps to distinguish between photochemical and vertical transport influences on the diurnal and seasonal cycles of HONO and NO2 given that columns are less sensitive than concentrations to the growth of the boundary layer.
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

8. OMI glyoxal retrieval algorithm
Original algorithm developed for GOME-2A in 2010 (Lerot et al., ACP).
Recently updated and adapted to OMI.
Relies on the standard DOAS approach.
Slant column retrieval
Fitting window: nm
Updated Absorption
cross-sections
CHOCHO +
O3, NO2 (two temperatures as in Alvarado et al., Miller et al., 2014), H2O, O4, Ring,
liquid water (fixed) and VRS
Conversion to vertical columns
Row-dependent background correction procedure (based on Pacific sector)
Air mass factors computed using a priori profile shapes provided by
the CTM IMAGESv2 over lands.
Airborne MAX-DOAS TORERO profile over oceans.
No cloud correction applied. Only pixels with cloud fraction < 20% are kept.
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

9. 3D-CTM IMAGES NMVOC species: 22 explicit Resolution: 2°x2.5°
Meteorological fields: ECMWF
Emission inventories:
Biogenic: MEGAN Anthropogenic: REAS (Regional Emission inventory in Asia)
Pyrogenic: GFEDv3.1
The model includes the chemistry of 22 precursor NMVOCs.
NMVOC species: explicit
IMAGES: Stavrakou et al., ACP, 2013
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

10. Time-series of CHOCHO VCD
Vertical column density
Good agreement between OMI (100km overpass) and MAX- DOAS
Smoothing of MAX- DOAS profiles with OMI AVKs improves the agreement between both data sets.
Observed seasonality can be mainly explained by larger biogenic and pyrogenic activities in summer
(x1E15 molec/cm2)
Unsmoothed MAX-DOAS
Smoothed MAX-DOAS
Additional pyrogenic and biogenic sources in summer and more efficient glyxoal precursors oxydation in summer.
In purple
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

11. Diurnal variation of CHOCHO
Vertical column density
Local time (h)
Near-surface concentration (0-200m layer)
Local time (h)
Diurnal variation of CHOCHO VCD seasonally averaged at both stations. The error bars correspond to the 1-sigma standard deviation.
Glyoxal VCDs at both stations agree within their combined error bars. However, we can see some differences between both stations: The dirunal variation is almost flat at Xianghe while a maximum is observed around noon in Beijing city center, except in winter but we have to be carefull with retrieval results in winter in Beijing because we have only data for one winter so the statisitics is poorer. Regarding the near-surface concentrations at both stations, they also agree within their combined error-bars and they show an increase in the early morning with a level-off around 10h LT, except in SON in Beijing (maximum at local noon as for the VCD) and in DJF (noisier data due to the lower statistics).
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

12. Comparison with 3D-CTM IMAGES
Vertical column density normalized WRT local noon
Vertical column density
· Simulations with the 3D-CTM IMAGES (Stavrakou et al., 2013) indicate a maximum around 13h-14h LT but also significantly lower VCD values. This underestimation could be due to the lack of knowledge about the all/exact formation mecanisms of glyoxal in heavily polluted environment as in the Beijing area. It could be also related to an underestimation of aromatics (BTX) emissions in the current anthropogenic emission inventories over China used by models (REAS in the present case; see Liu et al., 2012). If we multiply the aromatics emission by a factor of 6, we see that the agreement between MAXDOAS and the IMAGES model is significantly improved, especially in summer and in fall. Now, on a relative point of view, we see that the model is reproducing well the diurnal variation of CHOCHO VCD as illustrated on the left figure.
Liu et al., GRL, 2012: underestimation of aromatics (BTX) by a factor of 5-6 in the current anthropogenic emission inventories over China
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

13. Glyoxal : Formaldehyde ratio
CHOCHO:HCHO ratio RGF often used as an indicator of changes in VOC mixture:
Previous studies on this ratio:
Vrekoussis et al. (2010) – based on GOME-2:
RGF < 0.04 in urban areas
RGF > 0.06 in rural environments
Di Gangi et al. (2012) – based on in situ measurements during two campaigns:
RGF < 0.02 for biogenic regimes
RGF ~= for polluted air masses
Kaizer et al. (2015) – based on airborne in situ measurements:
RGF < for areas with high isoprenes emissions
RGF >0.03 for areas with high monoterpenes emissions
RGF can increase in case of local emissions (oil and gas production)
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

14. MAX-DOAS and OMI HCHO observations over Beijing/Xianghe
De Smedt et al., AMTD, 2015
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

15. Glyoxal : Formaldehyde ratio
CHOCHO/HCHO VCD ratio is ranging from to 0.055, except in DJF ( )
A good consistency is found with OMI
Higher RGF values are found during the winter where anthropogenic emissions are the largest
Results more consistent with Di Gangi and Kaizer than with Vrekoussis
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

16. Geometrical approximation vs OEM: CHOCHO
OEM SH=0.5km
OEM SH=1.0km
We Know that MAX-DOAS is mainly sensitive to the first two kilometers from the ground. And so using a higher SH is like adding a ghost column in the altitude range where MAX-DOAS has no sensitivity. and we see that the OEM estimation columns are shifted higher resulting in a better agreement with GA in summer.
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

17. Geometrical approximation vs OEM: HCHO
OEM SH=0.5km
OEM SH=1.0km
Such a good agreement is also obtained for NO2 in the visible.
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

18. Geometrical approximation vs OEM: CHOCHO
OEM SH=0.5km
OEM SH=1.0km
We Know that MAX-DOAS is mainly sensitive to the first two kilometers from the ground. And so using a higher SH is like adding a ghost column and we see that the OEM estimation columns are shifted higher resulting in a better agreement with GA in summer. We didn’t find any satisfactory explanation for this behaviour. We have performed a retrieval using a priori CHOCHO profiles from the TORERO campaign which took place over tropical Eastern Pacific Ocean. Of course the conditions differ strongly from the Beijing area but the idea was to test a different profile shape with a significant contribution of the free troposphere to the total column as illustatred by here by the plot of the profile shape.
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

19. Geometrical approximation vs OEM: CHOCHO
We Know that MAX-DOAS is mainly sensitive to the first two kilometers from the ground. And so using a higher SH is like adding a ghost column and we see that the OEM estimation columns are shifted higher resulting in a better agreement with GA in summer. We didn’t find any satisfactory explanation for this behaviour. We have performed a retrieval using a priori CHOCHO profiles from the TORERO campaign which took place over tropical Eastern Pacific Ocean. Of course the conditions differ strongly from the Beijing area but the idea was to test a different profile shape with a significant contribution of the free troposphere to the total column as illustatred by here by the plot of the profile shape. We see that taking a different a priori profile shape does not improve the agreement between oem and geometrical approximation. So this feature remains an open issue so far. Remains unexplained.
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

20. Concluding remarks
Tropospheric CHOCHO vertical profiles and corresponding column densities have been retrieved in the Beijing area by applying an OEM-based profiling tool to MAX-DOAS observations from June 2008 till December 2013
A good consistency is found between MAX-DOAS and OMI in terms of CHOCHO VCD seasonality (maximum in summer and a minimum in winter)
MAX-DOAS CHOCHO VCD has a maximum around local noon in the Beijing City centre, in reasonably good agreement with simulations from the 3D-CTM IMAGES; no marked diurnal variation is found in Xianghe
CHOCHO/HCHO VCD ratio is ranging from to 0.055, except in DJF ( ). A good consistency is found with OMI.
Further investigations:
Consistency between OEM and geometrical approximation
Comparison with the 3D-CTM IMAGES (surface concentration + different emission scenarios)
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

21. Thank you for your attention !
Acknowledgements
Belgian Federal Science Policy Office, Brussels (PRODEX and AGACC-II projects)
EU FP7 projects NORS and MarcoPolo
Thank you for your attention !
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

22. CHOCHO and HCHO diurnal variations
Vertical column density normalized WRT local noon
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

23. Geometrical approximation vs OEM: CHOCHO and HCHO at Bujumbura (Burundi)
OEM SH=1.0km
HCHO
OEM SH=1.0km
We found a similar discrepancy when using
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

24. Multiple scattering (O4) flag
Gielen et al., AMT, 2014
214.2
214.3
214.4
214.5
214.6
214.7
we opt to only study the
temporal variation of the measured O4 DSCDs. Strong temporal
variability due to enhanced multiple scattering commonly
occurs in optically thick clouds, and is seen less for
high aerosol optical depth
Fractional day
Strong temporal variability of O4 DSCDs due to enhanced multiple scattering commonly occurs in optically thick clouds, and is seen less for high aerosol optical depth
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium

25. OMI retrieval of CHOCHO VCDs
Original algorithm for GOME-2A developed in 2009 (Lerot et al., ACP, 2010)
Two steps: slant column retrieval (DOAS) + conversion of SCDs to VCDs
Algorithm updates:
Strong focus on GOME-2A/OMI consistency: Application of common settings to both sensors.
DOAS fit
Update of cross-section data base.
On-line convolution of cross-sections with a fitted slit function (for GOME-2A).
Daily mean background radiance spectrum as the reference instead of solar spectra.
Normalization procedure
Row- (OMI) or VZA-dependent (GOME-2) procedure
AMF computation
Over lands: A priori profiles provided by an updated version of the CTM IMAGES.
Over oceans: use of a fixed a priori profile (TORERO).
Extension of the weighting function LUT.
No cloud correction any more, only pixels with CF<0.2 are kept.  Reduction of the influence of the a priori information.
The total uncertainty is more likely underestimated since the total retrieval error includes only the smoothing error and retrieval noise and not the forward model parameter error. We are currently investigating the impact of parameters like aerosols and clouds.
7th International DOAS Workshop, 6-8 July 2015, Brussels, Belgium