Assuming only absorbing trace gas abundance and AOD are retrieved, using CO2 absorption band alone provides a DOF ~ 1.1, which is not enough to determine the two parameters. Combining O2 absorption band and assuming the O2 concentration is fixed at 0.2095, aerosol information can be obtained from the O2 band. DOF increases to 1.9, so retrieving CO2 and AOD simultaneously from normalized spectra is possible.

Figure 2. Degree of freedom (DOF) and information content (IC) of retrieval using CO2 band alone and combining CO2 and O2 bands.

CLARSReflected sunlight

Direct beam

LA basin

PBL height

Retrieval of CO2 Mixing Ratios from CLARS Measurements: Correcting Aerosol Induced Biases

Qiong Zhang1, Vijay Natraj2, Run-Lie Shia1, Coleen M. Roehl1, Yuk L. Yung1, and Stanley P. Sander2

1Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, USA.2Jet Propulsion Laboratory, California Institute of Technology, Pasadena, USA

Abstract

A Fourier transform spectrometer at the California Laboratory for Atmospheric Remote Sensing (CLARS) on the top of Mt Wilson, California, measures greenhouse gas concentrations in the Los Angeles basin using reflected sunlight. Observations include those with large viewing zenith angles (up to 83.1°), making the measurements very sensitive to aerosol scattering within the boundary layer. A previous study by the authors shows that ratioing the CO2 and O2 slant column densities (SCDs) can largely cancel the effect of aerosol scattering, but biases still exist due to the wavelength dependence of aerosol scattering. In this study, we show the possibility of correcting biases caused aerosols scattering. Preliminary results indicate that the information from CLARS-FTS spectra is not sufficient to constrain all the free parameters, including the aerosol SSA, aerosol optical depth, surface albedo, etc. In order to mitigate the influence of aerosol scattering, aerosol optical depth (AOD) is retrieved simultaneously with absorbing gas abundances. Both CO2 and O2 absorption bands are used in the retrieval. Aerosol information can be obtained in the O2 band assuming O2 concentration is fixed. By incorporating aerosol parameters into the CO2 and O2 retrievals, biases in caused by wavelength dependence of aerosol scattering can be considerably reduced.

CLARS measurement bias due to aerosol

Conclusions

Comparing with TCCON results

We demonstrate that combining O2 and CO2 absorption bands can obtain enough information to retrieve CO2 concentration and aerosol optical depth simultaneously.

Using both TCCON and CLARS Spectralon measurements, which are not affected by scattering, XCO2 along the CLARS West Pasadena measurement light path can be calculated with a simple box model. It is compared with our retrieval results.

Wavelength dependence of aerosol scattering typically causes a higher value of retrieved XCO2. By including aerosol parameters into the retrieval, the bias caused by aerosol scattering can be reduced.

TCCON XCO2 measurements (total column) are combined with CLARS Spectralon (free atmosphere) measurements to provide a comparison for CLARS West Pasadena measurements. We assume that CO2 is well-mixed in the boundary layer. This approach is insensitive to the boundary layer height. Compared with the calculated XCO2, CLARS measurements show a larger peak at 2:00 pm, one hour later than the peak in TCCON. Wind field and local emissions may explain the difference. The high value of CLARS XCO2 is partly due to aerosol scattering. When aerosol effects are considered, the retrieved XCO2 shows a lower peak.

DOF and IC analyses

CO2 absorption band alone

When pollution is heavy, we can see the 'U shape' as the concentrations of absorbing gas changes during the daytime. For the case of West Pasadena, measurement shows a low bias from the morning to the afternoon. The bias can be as large as 13% in the CO2 absorption band and 17% in the O2 absorption band (Figure 1). Biases in the 1.61m CO2 band are smaller than that in the 1.27m O2 band. Ratioing CO2 SCD by O2 SCD cannot cancel the bias.

CO2 and O2 absorption bands

Fit of spectra

Figure 1. Variations of the CLARS measurements from the morning to the afternoon, measured SCD vs. geometric SCD for (a) CO2 and (b) O2. A, B and C represent morning, noon and afternoon. Unit for CO2 SCD is scaled by 1022. Unit for O2 SCD is scaled by 1025.

Figure 3. Top panel shows the fit of measured spectra (red) and simulated spectra (blue) by the 2S-ESS model in the O2 absorption band. Radiances are normalized by the maximum value. Middle panel shows the residual of retrieval. Lower panel shows the probability density function (pdf) of the residual (red) and a Gaussian function curve (blue) generated to be compared with the residual pdf.

A two-stream-exact single scattering (2S-ESS) model [Spurr and Natraj, 2011] is used to fit the observed spectra. We retrieve CO2 H2O and AOD simultaneously. Local aerosol compositions are used. Measurement includes both CO2 and O2 bands. O2 is kept constant in the retrieval. Angstrom coefficient is obtained from AERONET station at Caltech [Holben et al., 1998]. RMS of the residual is comparable to the instrument SNR and matches the Gaussian distribution. CO2 band spectra show similar features.

0 0 1

0

1/ cos(sza) 1/ cos(vza)( )

1/ cos(sza)/ cos(sza) ( ) / cos(vza)

/ cos(sza)

TCC - SPCWP SPC

P P P

P

Figure 4. Schematic figure of CLARS measurement [Fu et al., 2014] .

References: Fu et al., AMT, 2014; Holben et al., Remote Sens. Environ., 1998; Spurr and Natraj, JQSRT, 2011; Wunch et al., Philos. T. R. Soc. A., 2011.

Figure 5. XCO2 measured by TCCON at Caltech, CLARS in West Pasadena and Spectralon. Black line shows the calculated XCO2 along the CLARS WP light path. Black marker-line shows XCO2 retrieved by a 2-stream model where the effect of aerosol is taken into account.