Constraining the vertical distribution of coastal dust aerosol using OCO-2 O2 A-band measurements
Introduction
The vertical distribution of atmospheric aerosols plays an important role in regulating the Earth's energy budget by scattering and absorbing sunlight (direct effect) and via aerosol-cloud interactions (indirect effect; IPCC, 2013; Zarzycki and Bond, 2010). In addition, the detection of vertical distribution of aerosols, which contribute the largest environmental risk as air pollutants, enables assessment of their impact on public health (Liu and Diner, 2017). Third, aerosol scattering effects provide one of the most important sources of uncertainty in greenhouse gas (GHG) retrievals from space in the near infrared (Kuang et al., 2002). A better knowledge of the aerosol vertical distribution, which affects the light path length in the gas absorption channels, is required to achieve high accuracy GHG retrievals (Crisp et al., 2008; Butz et al., 2009; O'Dell et al., 2018). Satellite (e.g., MODIS and MISR; Kahn et al., 2007) and ground-based (e.g., AERONET; Holben et al., 1998) measurements have been accurately and continuously monitoring the global column total aerosol optical depth. However, to date, limited information on the vertical structure of atmospheric aerosols has been obtained from passive remote sensing measurements. Lidar measurements (e.g., CALIPSO; Winker et al., 2010) provide more information about the aerosol vertical distribution, but lidar instruments have a narrow swath and therefore require weeks to observe just a fraction of a percent of the planet's surface area; it is therefore very difficult to obtain global coverage. This necessitates a new era of research pursuing alternative approaches to solve this challenging problem.
The use of oxygen (O2) absorption measurements to constrain cloud and aerosol profiles was first proposed by Yamamoto and Wark (1961). The physical basis is that (1) O2 is uniformly distributed in the atmosphere with a mixing ratio of ~0.20955, (2) its spectrally-dependent absorption cross sections are reasonably well known (Drouin et al., 2017), and (3) as aerosols and clouds scatter light back to space, they leave distinctive signatures in different parts of the observed O2 spectra, which are associated with the column aerosol/cloud optical depth and vertical structure. By studying these signatures, we can quantify the vertical distribution of aerosols and clouds. This technique has been successfully applied to quantify the cloud top pressure and cloud thickness (e.g., O'Brien and Mitchell, 1992; Heidinger and Stephens, 2000; Richardson et al., 2017), quantify the impact of aerosol scattering on dry air mass used in CO2 and CH4 retrievals (Bösch et al., 2006, 2011; Geddes and Bösch, 2015; O'Dell et al., 2018; Wu et al., 2018) and to characterize the impact of the 3-D structure of clouds on CO2 retrievals (Massie et al., 2017). However, with respect to constraining the atmospheric aerosol vertical distribution, the majority of studies are still in the theoretical phase (e.g., Hollstein and Fischer, 2014; Sanders et al., 2015; Colosimo et al., 2016; Ding et al., 2016; Davis and Kalashnikova, 2019; Hou et al., 2017), with very few applications using real remote sensing measurements (e.g., Dubuisson et al., 2009; Sanghavi et al., 2012; Xu et al., 2017b; Zeng et al., 2018; Nanda et al., 2019).
Here we propose the use of a spectral sorting approach (Liou, 2002; Richardson et al., 2017; Zeng et al., 2018) for retrieving the total aerosol optical depth (AOD) and aerosol layer height (ALH) using hyperspectral O2 A-band measurements from Orbiting Carbon Observatory-2 (OCO-2; Crisp et al., 2008), a NASA mission dedicated to measuring the concentration of atmospheric carbon dioxide (CO2). This approach has been successfully applied to O2 absorption measurements from the California Laboratory of Atmospheric Remote Sensing (CLARS; Zeng et al., 2017; He et al., 2019), a mountain-top Fourier Transform Spectrometer (FTS) overlooking the Los Angeles megacity, to detect the aerosol loading in the boundary layer. In this study, the effectiveness of this approach for satellite-based measurements is evaluated, using OCO-2 as a test case. We selected the western Sahara coast as the study area to minimize effects from variable land surface reflectance and to maximize the variability of aerosol layer height due to frequent dust storms over this coastal region. The development of an aerosol profile retrieval technique for passive remote sensing observations will be useful for future missions to produce maps of aerosol vertical structure on a global scale. In addition, the aerosol vertical structure constrained by this technique may provide a more efficient method for minimizing aerosol-related biases in CO2 and CH4 retrievals.
In Section 2, we describe how we obtain collocated measurements from OCO-2 and CALIPSO over the western Sahara coast. Section 3 illustrates the sensitivity of the O2 A-band to aerosol vertical structure using simulations of OCO-2 measurements. In Section 4, the AOD and ALH are retrieved using look-up tables (LUTs). The results are discussed in Section 5, and conclusions provided in Section 6.
Section snippets
Western Sahara coast
The coastline of Mauritania in the western Sahara Desert faces the Atlantic Ocean (Fig. 1). Each year, dust storms bring in hundreds of millions of tons of dust from the desert to the Atlantic Ocean (Prospero and Mayo-Bracero, 2013). Some of the dust reaches North and South America and affects the local air quality as well as climate, soil fertility, marine biology at large scales. A better knowledge of the dust vertical structure will improve our understanding of dust transport and assessment
Sensitivity of O2 A-band measurement to aerosol vertical structure
In this section, the OCO-2 forward model, which is used to simulate the observed radiance and construct look up tables (LUTs; Section 4) for different AODs and ALHs, is introduced. The radiance sensitivity, which is the response of the radiance to different aerosol vertical structures, is further investigated using the forward model. It is expected that the channels with higher sensitivity will provide better constraints on the aerosol vertical structure.
A spectral sorting approach for constraining aerosol profile
A spectral sorting approach has been successfully applied to O2 1Δ band measurements at 1.27 μm from the CLARS-FTS instrument to profile aerosols within the planetary boundary layer in the Los Angeles megacity (Zeng et al., 2018). As shown in Fig. 7, the retrieval approach is implemented by first constructing look up tables (LUTs) with different AODs and ALHs using the OCO-2 forward model. Due to changes in observation geometry and atmospheric conditions on a daily basis, the LUTs were built as
Challenges in applying spectral sorting approach to measurements over land
There are two key challenges when trying to apply our spectral sorting approach to satellite measurements of the O2 A-band over land. The first is the determination of surface reflectance. The surface reflectance in the O2 A-band over land has a large range of variability for different land surface types (Moody et al., 2005). Especially over urban regions, the main source of anthropogenic aerosol emissions, the complex landscape makes the BRDF very challenging to model. Any error in surface
Conclusions
We describe a spectral sorting approach for constraining aerosol optical depth and vertical structure using O2 A-band measurements from OCO-2. The effectiveness of the approach is demonstrated by application to dusty soundings over the western Sahara coast and comparison with co-located lidar measurements from CALIPSO CALIOP. Using the OCO-2 forward model to emulate OCO-2 measurements, we find that the observed O2 A-band hyperspectral measurements have high sensitivity to the aerosol vertical
Declaration of competing interest
None.
Acknowledgement
We thank Run-Lie Shia at Caltech, Suniti Sanghavi at JPL, and Chao Liu at NUIST for stimulating discussions. The OCO-2 Forward model is available at https://github.com/nasa/RtRetrievalFramework. The L1bSc OCO-2 radiances are available online from the NASA Goddard GES DISC at https://disc. gsfc.nasa.gov/datacollection/OCO2_ L1B_Science_7.html. MERRAero monthly 3-h averaged dust column density data can be downloaded from (//portal.nccs.nasa.gov/cgi-lats4d/webform.cgi%3f%26i=GEOS-5/MERRAero/monthly/tavg3hr_2d_aer_Nx)
References (57)
- et al.
Evaluation of the surface PM2.5 in version 1 of the NASA MERRA aerosol reanalysis over the United States
Atmos. Environ.
(2016) - et al.
Multispectrum analysis of the oxygen A-band
J. Quant. Spectrosc. Radiat. Transf.
(2017) - et al.
Estimating the altitude of aerosol plumes over the ocean from reflectance ratio measurements in the O2 A-band
Remote Sens. Environ.
(2009) - et al.
Aeronet – a federated instrument network and data archive for aerosol characterization
Remote Sens. Environ.
(1998) - et al.
An algorithm for hyperspectral remote sensing of aerosols: 2. Information content analysis for aerosol parameters and principal components of surface spectra
J. Quant. Spectrosc. Radiat. Transf.
(2017) - et al.
A fast linearized pseudo-spherical two orders of scattering model to account for polarization in vertically inhomogeneous scattering–absorbing media
J. Quant. Spectrosc. Radiat. Transf.
(2007) - et al.
First operational BRDF, albedo nadir reflectance products from MODIS
Remote Sens. Environ.
(2002) Simultaneous derivation of intensities and weighting functions in a general pseudo-spherical discrete ordinate radiative transfer treatment
J. Quant. Spectrosc. Radiat. Transf.
(2002)VLIDORT: a linearized pseudo-spherical vector discrete ordinate radiative transfer code for forward model and retrieval studies in multilayer multiple scattering media
J. Quant. Spectrosc. Radiat. Transf.
(2006)- et al.
TROPOMI on the ESA Sentinel-5 Precursor: a GMES mission for global observations of the atmospheric composition for climate, air quality and ozone layer applications
Remote Sens. Environ.
(2012)