An algorithm for hyperspectral remote sensing of aerosols: 3. Application to the GEO-TASO data in KORUS-AQ field campaign
Introduction
Aerosols are an important component of the global atmosphere. They are composed of liquid and solid particles suspended in the air and can originate from either natural or manmade sources [1], [2]. By scattering and absorbing solar radiation and further modifying the properties of clouds, aerosols have a significant impact on climate change, air pollution, visibility and the ecological environment [3], [4], [5]. In the past decades, polar-orbiting satellites have demonstrated the ability to observe aerosols and other pollutants (such as SO2 and NO2) affecting air quality [6], [7], [8], [9], [10], [11], [12], [13], [14]. However, the impact of polar-orbiting observations has been limited by their coarse temporal resolution (usually once per day) with respect to change of aerosol source distributions that have distinct diurnal cycle (such as urban emissions) or are sporadic (such as fires and dust). This limitation is insufficient to observe the details of the diurnal nature of air pollution events that can develop over timescales of an hour to within a day [7], [15], [16]. Conversely, observations of geostationary (GEO) satellites can overcome these problems and provide observations many times throughout the day [17], [18], [19], [20], [21], [22], [23]. Recently, an international effort has been developed to launch a constellation of GEO satellite instruments focused on air quality over Asia, North America and Europe, including the Geostationary Environment Monitoring Spectrometer (GEMS) [24], [25], Tropospheric Emissions: Monitoring of Pollution (TEMPO) [26], [27] and Sentinel-4 [28] missions, respectively. However, while GEO sensors can have an advantage in temporal sampling, they do suffer from reduced spatial resolution and must cope with increased challenges associated with viewing geometry and scattering angles.
This paper presents the third part of a series of studies that aim to develop hyperspectral remote sensing techniques for aerosol retrievals from a newly developed instrument called Geostationary Trace gas and Aerosol Sensor Optimization (GEO-TASO) [29], [30]. As the airborne version of the upcoming air quality satellite instruments onboard GEMS and TEMPO, GEO-TASO measures hyperspectral backscattered ultraviolet (UV), visible (VIS) and near-infrared (NIR) radiation. TEMPO has been selected as the first Earth Venture Instrument by the National Aeronautics and Space Administration (NASA), and will be launched in 2022 to measure atmospheric pollution over North America from space, by using the hyperspectral UV and VIS spectroscopy hourly and with the spatial resolution about 4 × 4 km2 [26]. Meanwhile, to improve the use of satellites to monitor air quality for public health and benefit, several field campaigns have been carried out by NASA, such as the Deriving Information on Surface Conditions form Column and Vertically Resolved Observations Relevant to Air Quality (DISCOVER-AQ) in Houston and Denver during 2013–2014 [29], [31], and the Korea-United Sated Air Quality Study (KORUS-AQ) in Korea during 2016 [32], [33]. From the DISCOVER-AQ and KORUS-AQ campaigns, a series of airborne hyperspectral observation of GEO-TASO have been obtained and a three-part of study was carried out to explore and develop algorithms to retrieve aerosols from the real hyperspectral data in support of missions such as TEMPO.
In the first two parts of this study [34], [35], we have developed a theoretical framework to retrieve aerosol and surface properties simultaneously from hyperspectral measurements in the visible spectrum. Here, we apply that theoretical algorithm to the observation data collected by GEO-TASO during KORUS-AQ. Our retrieval is based on an optimal-estimation (OE) approach to conduct the spectral fitting with inline radiative transfer, because the high dimensionality of state vector with multiple retrieved parameters may led to the much too large lookup-tables (LUTs). For computational reasons, the retrievals use the inline forward model instead of the LUTs approach [9], [36] in this study. This strategy is consistent with the past studies that have applied OE method to invert aerosol properties from a wide range of optical instruments onboard either space-borne or airborne platforms, including the SeaWiFS [37], SEVIRI [38], [39], [40], [41], OMI [42], SAG II [43], IASI [44], AASTR [45], [46], MISR [47], POLDER/PARASOL [48], [49], [50], [51], CAPI [52], [53], DPC [54], PSAC [55], GEMS [24], EPIC/DSCOVR [56], RSP [57], [58], [59], [60], [61], AirMSPI [62], [63], [64], SPEX airborne/ACEPOL [65], [66], APEX/ESA [67], AVRIS-NG [68], [69] and GEO-TASO [34], [35]. The definition of these sensors’ acronyms is listed in Table 1, so are the names of the surface models, the forward models, and the OE models used by these past studies to implement OE-based inversion. The Levenberg-Marquardt method and Quasi-Newton methods are usually used for optimization in the framework of the inversion theory [70].
For the hyperspectral remote sensing, the OE approaches have been already used in the retrieval of gas concentration [71], [72] and atmospheric correction of hyperspectral imaging [68], [69] for AVRIS-NG. However, challenges in these past studies remain regarding the retrieval of aerosol parameters, especially in the treatment of surface reflectance. In the inversion, since the accuracy of surface reflectance estimation can directly affect the separation of the atmospheric contribution of radiances (or path radiance) from the surface contribution to the hyperspectral measurements at altitude of satellite or aircraft, one major source of uncertainty in the algorithm of aerosol retrievals arise from the approach to characterize the hyperspectral surface reflectance. To tackle this issue, our algorithm framework as shown in the first two-parts of this study [34], [35], is to simultaneously retrieve the aerosol and surface properties. The unique part of this framework is to decompose the surface reflectance spectra into different principal components (PCs). Thus we only need to retrieve several weighting coefficients of PCs instead of the reflectance band-by-band for full spectra [34]. Indeed, for the hyperspectral instruments like GEO-TASO, they have tens to hundreds of bands in the ultraviolet (UV) and visible (VIS) spectrum and a PC-based characterization of surface reflectance in these bands are not only computationally appealing but also physically sound because the reflectance in these bands are often highly correlated. Hou et al. [35] also showed that the PCs of surface spectra in the atmospheric window channel could be approximately derived from the top-of the atmosphere reflectance in the cases of the low AOD. Here, to investigate the feasibility of the OE-based retrieval framework with principal component analysis (PCA), we apply its algorithm theoretical basis (as described in the part 1 and part 2 of this study [34], [35]) to conduct the retrievals of aerosol and surface properties from the real data, i.e., the hyperspectral measurements of GEO-TASO during the KORUS-AQ campaign. For completeness, we briefly describe the methodology of our OE-based inversion in Section 2, and then present the GEO-TASO data processing and preparation in Section 3. Afterward, the retrieval results are investigated and validated in Section 4. Finally, the summary and collusion are provided in Section 5.
Section snippets
Methodology
Fig. 1 shows the flowchart of our algorithm. The algorithm is based on the OE inversion framework described in Section 2.1. Because of the fine spectral resolution (~0.28 nm) of the GEO-TASO, to expedite the inversion, GEO-TASO data thinning is conducted through band selection method in Section 2.2. To reconstruct the spectral surface reflectance, as an initial guess, the common PCs for various surface reflectance spectra are first extracted from the spectral reflectance libraries for different
Data processing and preparation
GEO-TASO data in KORUS-AQ field campaign are processed to suit two purposes. The first is to imitate the geostationary spatial distribution of TEMPO by focusing on the observations of different flights in the same study area, while the second is to analyze observations over 7 AERONET sites to evaluate the retrieved properties of aerosols. Meanwhile, a prior aerosol model has been extracted based on the AERONET to constrain the inversion in KORUS-AQ field campaign.
OE retrieval results in the study area
Fig. 8 illustrates the retrieved surface reflectance at the initial step, the adjusted surface reflectance, as well as the updated PCs (for 20 selected bands) and corresponding prior constraint of weighting coefficients for 3 different surface types. We carried out the initial retrieval process with the common PCs (flag = 1) based on surface types classified in Fig. 4(g), and consequently, the retrieved surface reflectance for these 3 different surface types were obtained and are shown in panel
Summary and conclusion
As the third part of a series of studies to retrieve aerosol properties form the hyperspectral radiance measurements by new instrument GEO-TASO and future geostationary satellite TEMPO, we developed and conducted the PC-based OE algorithm to retrieve spectral AODs from the GEO-TASO data collected in the KORUS-AQ field campaign. We can summarize our work as follows.
- (1)
With the combination of the band selection method presented in our previous work [35] and the real data from GEO-TASO, those best
CRediT authorship contribution statement
Weizhen Hou: Methodology, Formal analysis, Writing - original draft. Jun Wang: Conceptualization, Funding acquisition, Supervision, Methodology, Formal analysis, Writing - original draft. Xiaoguang Xu: Software, Validation, Visualization. Jeffrey S. Reid: Writing - review & editing. Scott J. Janz: Data curation. James W. Leitch: Data curation.
Declaration of Competing Interest
The authors declare that they have no known competing financial interestsor personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
Funding for this study was provided by the NASA Earth Science Division as part of NASA's GEOCAPE Decadal Mission study and the KORUS-AQ field study as well as Office of Naval Research (ONR's) Multidisciplinary University Research Initiatives (MURI) Program under the award N00014-16-1-2040. J. Wang also acknowledges TEMPO mission (grant SV7-87011 managed by Harvard-Smithsonian Center for Astrophysics) for partial support and Dr. Jassim A. Al-saadi for his leadership for GEOCAPE study. The
Reference (96)
- et al.
Monitoring of aerosol forcing of climate from space: analysis of measurement requirements
J Quant Spectrosc Ra
(2004) - et al.
Aerosol remote sensing over land: a comparison of satellite retrievals using different algorithms and instruments
Atmos Res
(2007) - et al.
Past, present, and future of global aerosol climatologies derived from satellite observations: a perspective
J Quant Spectrosc Ra
(2007) - et al.
Toward unified satellite climatology of aerosol properties
J Quant Spectrosc Ra
(2010) - et al.
Estimation of surface NO2 volume mixing ratio in four metropolitan cities in korea using multiple regression models with OMI and AIRS Data
Remote Sens (Basel)
(2017) - et al.
Potential application of VIIRS Day/Night Band for monitoring nighttime surface PM2.5 air quality from space
Atmos Environ
(2016) - et al.
Improved algorithm for MODIS satellite retrievals of aerosol optical thickness over land in dusty atmosphere: implications for air quality monitoring in China
Remote Sens Environ
(2010) - et al.
Polarimetric remote sensing of atmospheric aerosols: instruments, methodologies, results, and perspectives
J Quant Spectrosc Ra
(2019) - et al.
Space-based remote sensing of atmospheric aerosols: the multi-angle spectro-polarimetric frontier
Earth-sci Rev
(2015) - et al.
Observations of residual submicron fine aerosol particles related to cloud and fog processing during a major pollution event in Beijing
Atmos Environ
(2014)
A numerical testbed for remote sensing of aerosols, and its demonstration for evaluating retrieval synergy from a geostationary satellite constellation of GEO-CAPE and GOES-R
J Quant Spectrosc Ra
Tropospheric emissions: monitoring of pollution (TEMPO)
J Quant Spectrosc Ra
An algorithm for hyperspectral remote sensing of aerosols: 1. Development of theoretical framework
J Quant Spectrosc Ra
An algorithm for hyperspectral remote sensing of aerosols: 2. Information content analysis for aerosol parameters and principal components of surface spectra
J Quant Spectrosc Ra
Use of MODIS-derived surface reflectance data in the ORAC-AATSR aerosol retrieval algorithm: impact of differences between sensor spectral response functions
Remote Sens Environ
An optimization approach for aerosol retrievals using simulated MISR radiances
Atmos Res
Angular dependence of aerosol information content in CAPI/TanSat observation over land: effect of polarization and synergy with A-train satellites
Remote Sens Environ
Aerosol retrieval sensitivity and error analysis for the cloud and aerosol polarimetric imager on board TanSat: the effect of multi-angle measurement
Remote Sens (Basel)
Directional Polarimetric Camera (DPC): monitoring aerosol spectral optical properties over land from satellite observation
J Quant Spectrosc Ra
Cloud information content in EPIC/DSCOVR’s oxygen A- and B-band channels: an optimal estimation approach
J Quant Spectrosc Ra
Optimal estimation for imaging spectrometer atmospheric correction
Remote Sens Environ
Optimal estimation of spectral surface reflectance in challenging atmospheres
Remote Sens Environ
Mapping methane concentrations from a controlled release experiment using the next generation airborne visible/infrared imaging spectrometer (AVIRIS-NG)
Remote Sens Environ
A decomposition method for large-scale box constrained optimization
Appl Math Comput
AERONET—A federated instrument network and data archive for aerosol characterization
Remote Sens Environ
A novel approach to hyperspectral band selection based on spectral shape similarity analysis and fast branch and bound search
Eng Appl Artif Intel
A pilot study of shortwave spectral fingerprints of smoke aerosols above liquid clouds
J Quant Spectrosc Ra
The ASTER spectral library version 2.0
Remote Sens Environ
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 Ra
The modern aerosol retrieval algorithms based on the simultaneous measurements of the intensity and polarization of reflected solar light: a review
Frontiers in Environmental Science
Principal component analysis of remote sensing of aerosols over oceans
IEEE T Geosci Remote
A satellite view of aerosols in the climate system
Nature
Climate change 2014: synthesis report
A review and framework for the evaluation of pixel-level uncertainty estimates in satellite aerosol remote sensing
Atmos Meas Tech
Application of OMI, SCIAMACHY, and GOME-2 satellite SO2retrievals for detection of large emission sources
J Geophys Res: Atmos
Geostationary satellite retrievals of aerosol optical thickness during ACE-Asia
J Geophys Res: Atmos
GOES 8 retrieval of dust aerosol optical thickness over the Atlantic Ocean during PRIDE
J Geophys Res: Atmos
High temporal resolution aerosol retrieval using Geostationary Ocean Color Imager: application and initial validation
J Appl Remote Sens
Aerosol optical depth (AOD) retrieval using simultaneous GOES-East and GOES-West reflected radiances over the western United States
Atmos Meas Tech
Constraints on methane emissions in North America from future geostationary remote-sensing measurements
Atmos Chem Phys
GOCI Yonsei Aerosol Retrieval (YAER) algorithm and validation during the DRAGON-NE Asia 2012 campaign
Atmos Meas Tech
Optimal estimation-based algorithm to retrieve aerosol optical properties for GEMS measurements over Asia
Remote Sens (Basel)
Evaluation of ozone profile and tropospheric ozone retrievals from GEMS and OMI spectra
Atmos Meas Tech
Quantification and mitigation of the impact of scene inhomogeneity on Sentinel-4 UVN UV-VIS retrievals
Atmos Meas Tech
Nitrogen dioxide observations from the Geostationary Trace gas and Aerosol Sensor Optimization (GeoTASO) airborne instrument: retrieval algorithm and measurements during DISCOVER-AQ Texas 2013
Atmos Meas Tech
The GeoTASO airborne spectrometer project. Proc. SPIE 9218
Earth Observ Syst XIX
Factors that influence surface PM2.5 values inferred from satellite observations: perspective gained for the US Baltimore–Washington metropolitan area during DISCOVER-AQ
Atmos Chem Phys
Cited by (22)
The polarization crossfire (PCF) sensor suite focusing on satellite remote sensing of fine particulate matter PM<inf>2.5</inf> from space
2022, Journal of Quantitative Spectroscopy and Radiative TransferCitation Excerpt :In fact, the remote sensing capability of PM estimates largely depends on the aerosol key parameters detected by satellites [14–18]. At present, the satellite products of aerosol optical depth (AOD) are relatively mature [1,19,20–27], but the inversion accuracy of fine particles by satellite still needs to be improved. Therefore, the improvement of the detection capability of satellite sensors plays an important role in the remote sensing estimation of PM mass concentration.
Spectral Reconstruction Method of Mid-Infrared Surface Characteristics Based on Non-Negative Matrix Factorization
2024, Guang Pu Xue Yu Guang Pu Fen Xi/Spectroscopy and Spectral AnalysisHigh-Resolution Aerosol Optical Depth Retrieval in Urban Areas Based on Sentinel-2
2023, Photogrammetric Engineering and Remote SensingDetection of counterfeit holograms using hyperspectral imaging
2023, Proceedings of SPIE - The International Society for Optical EngineeringOptimal estimation algorithm research for aerosol fine-mode fraction retrieval from polarimetric measurements
2022, National Remote Sensing Bulletin