An algorithm for hyperspectral remote sensing of aerosols: 3. Application to the GEO-TASO data in KORUS-AQ field campaign

https://doi.org/10.1016/j.jqsrt.2020.107161Get rights and content

Highlights

  • An algorithm is prototyped for TEMPO to retrieve aerosols & surfaces simultaneously.

  • The prototyping is implemented by applying GEO-TASO data in KORUS-AQ to mimic TEMPO.

  • Spectral AOD retrievals from GEO-TASO are validated with AERONET data.

  • Principal components of surface reflectance are retrieved iteratively together with AOD.

  • Next-step applications of this algorithm for TEMPO instruments are discussed.

Abstract

This paper describes the third part of a series of investigations to develop algorithms for simultaneous retrieval of aerosol parameters and surface spectral reflectance from GEOstationary Trace gas and Aerosol Sensor Optimization (GEO-TASO) instrument. Since the algorithm is designed for future hyperspectral and geostationary satellite sensors, such as Tropospheric Emissions: Monitoring of Pollution (TEMPO), it is applied to GEO-TASO data measured over the same area by different flights as part of the Korea-United Stated Air Quality Study (KORUS-AQ) field campaign in 2016. While GEO-TASO has a spectral sampling interval of ~0.28 nm in the visible, its data is thinned through a band selection approach with consideration of atmospheric transmittance and different surface types, which yields 20 common spectral bands to be used by the algorithm. The algorithm starts with 4 common principal components (PCs) for surface spectral reflectance extracted from various spectral libraries; constraints of surface reflectance and aerosol model parameters are obtained respectively from k-means clustering analysis of the Rayleigh-corrected GEO-TASO spectra and AERONET data. The algorithm then proceeds iteratively with an optimal estimation approach to update PCs and retrieve aerosol optical depth (AOD) from GEO-TASO measured spectra until state vector converges. The comparison of AODs between GEO-TASO retrievals (y) and 7 AERONET (x) sites reveals that the iterative updates of surface PCs (and so surface reflectance) improve the inversions of fine-mode AOD, fine-mode fraction of AOD, Ångström exponent, and AOD at all (440, 550, 550, 675 nm) wavelengths. At 440 nm, the linear fitting equation, the Pearson correlation coefficient (R2), and mean absolute error are improved respectively from y = 0.72x + 0.11, 0.53, and 0.05 (without update of PCs) to y = 1.055x + 0.01, 0.76, and 0.033. Future work is to prepare the algorithm for TEMPO that carries an enhanced version of GEO-TASO instrument.

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 GEOsingle bondCAPE 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 GEOsingle bondCAPE study. The

Reference (96)

  • J. Wang et al.

    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

    (2014)
  • P. Zoogman et al.

    Tropospheric emissions: monitoring of pollution (TEMPO)

    J Quant Spectrosc Ra

    (2017)
  • W. Hou et al.

    An algorithm for hyperspectral remote sensing of aerosols: 1. Development of theoretical framework

    J Quant Spectrosc Ra

    (2016)
  • W. Hou 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 Ra

    (2017)
  • A.M. Sayer et al.

    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

    (2012)
  • D.J. Diner et al.

    An optimization approach for aerosol retrievals using simulated MISR radiances

    Atmos Res

    (2012)
  • X. Chen et al.

    Angular dependence of aerosol information content in CAPI/TanSat observation over land: effect of polarization and synergy with A-train satellites

    Remote Sens Environ

    (2017)
  • X. Chen et al.

    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)

    (2017)
  • Z. Li et al.

    Directional Polarimetric Camera (DPC): monitoring aerosol spectral optical properties over land from satellite observation

    J Quant Spectrosc Ra

    (2018)
  • A.B. Davis et al.

    Cloud information content in EPIC/DSCOVR’s oxygen A- and B-band channels: an optimal estimation approach

    J Quant Spectrosc Ra

    (2018)
  • D.R. Thompson et al.

    Optimal estimation for imaging spectrometer atmospheric correction

    Remote Sens Environ

    (2018)
  • D.R. Thompson et al.

    Optimal estimation of spectral surface reflectance in challenging atmospheres

    Remote Sens Environ

    (2019)
  • A.K. Thorpe et al.

    Mapping methane concentrations from a controlled release experiment using the next generation airborne visible/infrared imaging spectrometer (AVIRIS-NG)

    Remote Sens Environ

    (2016)
  • J. Yu et al.

    A decomposition method for large-scale box constrained optimization

    Appl Math Comput

    (2014)
  • B.N. Holben et al.

    AERONET—A federated instrument network and data archive for aerosol characterization

    Remote Sens Environ

    (1998)
  • S. Li et al.

    A novel approach to hyperspectral band selection based on spectral shape similarity analysis and fast branch and bound search

    Eng Appl Artif Intel

    (2014)
  • X. Xu et al.

    A pilot study of shortwave spectral fingerprints of smoke aerosols above liquid clouds

    J Quant Spectrosc Ra

    (2018)
  • A.M. Baldridge et al.

    The ASTER spectral library version 2.0

    Remote Sens Environ

    (2009)
  • R. Spurr

    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

    (2006)
  • A.A. Kokhanovsky

    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

    (2015)
  • V. Zubko et al.

    Principal component analysis of remote sensing of aerosols over oceans

    IEEE T Geosci Remote

    (2007)
  • Y.J. Kaufman et al.

    A satellite view of aerosols in the climate system

    Nature

    (2002)
  • Climate change 2014: synthesis report

    (2014)
  • A.M. Sayer et al.

    A review and framework for the evaluation of pixel-level uncertainty estimates in satellite aerosol remote sensing

    Atmos Meas Tech

    (2020)
  • V.E. Fioletov et al.

    Application of OMI, SCIAMACHY, and GOME-2 satellite SO2retrievals for detection of large emission sources

    J Geophys Res: Atmos

    (2013)
  • J. Wang et al.

    Geostationary satellite retrievals of aerosol optical thickness during ACE-Asia

    J Geophys Res: Atmos

    (2003)
  • J. Wang et al.

    GOES 8 retrieval of dust aerosol optical thickness over the Atlantic Ocean during PRIDE

    J Geophys Res: Atmos

    (2003)
  • Y. Zhang et al.

    High temporal resolution aerosol retrieval using Geostationary Ocean Color Imager: application and initial validation

    J Appl Remote Sens

    (2014)
  • H. Zhang et al.

    Aerosol optical depth (AOD) retrieval using simultaneous GOES-East and GOES-West reflected radiances over the western United States

    Atmos Meas Tech

    (2013)
  • N. Bousserez et al.

    Constraints on methane emissions in North America from future geostationary remote-sensing measurements

    Atmos Chem Phys

    (2016)
  • M. Choi et al.

    GOCI Yonsei Aerosol Retrieval (YAER) algorithm and validation during the DRAGON-NE Asia 2012 campaign

    Atmos Meas Tech

    (2016)
  • M. Kim et al.

    Optimal estimation-based algorithm to retrieve aerosol optical properties for GEMS measurements over Asia

    Remote Sens (Basel)

    (2018)
  • J. Bak et al.

    Evaluation of ozone profile and tropospheric ozone retrievals from GEMS and OMI spectra

    Atmos Meas Tech

    (2013)
  • Chance K., Liu X., Suleiman R.M., Flittner D.E., Al-Saadi J., Janz S.J. Tropospheric emissions: monitoring of pollution...
  • S. Noël et al.

    Quantification and mitigation of the impact of scene inhomogeneity on Sentinel-4 UVN UV-VIS retrievals

    Atmos Meas Tech

    (2012)
  • C.R. Nowlan et al.

    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

    (2016)
  • J.W. Leitch et al.

    The GeoTASO airborne spectrometer project. Proc. SPIE 9218

    Earth Observ Syst XIX

    (2014)
  • S. Crumeyrolle et al.

    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

    (2014)
  • 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 Transfer
      Citation 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.

    • Detection of counterfeit holograms using hyperspectral imaging

      2023, Proceedings of SPIE - The International Society for Optical Engineering
    View all citing articles on Scopus
    View full text