TL;DR
This paper presents a new, globally gridded dataset of aerosol composition derived from AERONET sun photometer measurements using the GRASP/component algorithm. This dataset offers a significant advancement in understanding aerosol impacts on climate by providing long-term, value-added information on key aerosol types, and it reveals important insights into the performance of the MERRA-2 reanalysis product.
What Problem Does This Paper Address?
Aerosols are a major source of uncertainty in climate change assessments due to their complex interactions with radiation. While ground-based networks like AERONET have provided decades of aerosol optical property data, directly inferring aerosol composition (i.e., the mix of different aerosol types like black carbon, dust, sulfates, etc.) from these measurements has been challenging. Existing reanalysis products like MERRA-2 offer global aerosol composition estimates, but their validation has been limited, particularly for the entire atmospheric column. This paper addresses the need for a robust, globally consistent, and long-term dataset of aerosol composition to improve our understanding of aerosol radiative forcing and to validate/improve climate models.
Methods
The core of this work is the application of the Generalized Retrieval of Atmosphere and Surface Properties (GRASP) algorithm with a component inversion approach. Unlike traditional methods that first retrieve optical properties and then infer composition, GRASP/component directly retrieves the mass concentrations of different aerosol components from AERONET's directional sky radiances and spectral aerosol optical depth (AOD). This approach assumes a mixture of known aerosol components (e.g., black carbon, brown carbon, dust, sulfates) with fixed complex refractive indices, and uses the Maxwell-Garnet mixing rule to model the overall aerosol optical properties. The authors also performed extensive validation of their retrieved aerosol optical properties against AERONET standard products and compared their composition retrievals with MERRA-2 products.
Datasets
- AERONET (Aerosol Robotic Network) measurements: This study utilized Level 1.5 and Level 2.0 data from AERONET sun photometer measurements, including directional sky radiances and spectral AOD, from 918 sites globally.
- MERRA-2 (Modern-Era Retrospective Analysis for Research and Applications, version 2) products: Hourly, 3-hourly, and monthly data for black carbon (BC) and dust column mass concentrations, as well as total and speciated extinction and scattering AOD at 550 nm, were used for comparison.
A new value-added, long-term aerosol composition dataset derived from AERONET is also made available online.
Key Insights and Ideas
- Direct retrieval of composition is feasible and valuable: The GRASP/component approach demonstrates that it's possible to directly infer aerosol composition from ground-based remote sensing data, bypassing intermediate steps and potentially reducing uncertainties.
- AERONET data holds rich compositional information: The study highlights that decades of AERONET data, when processed with advanced algorithms, can provide a global, long-term record of aerosol composition.
- MERRA-2 shows promise but needs refinement: The comparison with MERRA-2 reveals that while MERRA-2 captures similar temporal trends for BC and dust, it often underestimates their concentrations, particularly in certain regions. This suggests areas where MERRA-2's aerosol modeling or assimilation might need improvement.
- Regional variations are significant: The dataset reveals distinct spatial patterns for different aerosol components, with high BC concentrations in Southeast Asia, South Asia, and Africa, and significant dust contributions in arid regions like the Sahara Desert.
Key Results
- Aerosol Optical Property Validation: GRASP/component retrievals of aerosol optical properties (AOD, AAOD, SSA, AE) showed excellent agreement with AERONET standard products (e.g., R=0.965 for AAOD, R=0.951 for SSA).
- Global Aerosol Composition Dataset: A new, value-added, long-term global dataset of aerosol composition (including BC, brown carbon, iron oxides, organic carbon, quartz, and inorganic salts) was generated from 918 AERONET sites.
- MERRA-2 Comparison (BC): GRASP/AERONET BC retrievals showed good temporal correlation with MERRA-2 (R=0.60-0.85) but often indicated significantly higher concentrations (up to 142% difference in sub-Sahel).
- MERRA-2 Comparison (Dust): GRASP/AERONET dust retrievals also correlated well with MERRA-2 (R=0.75-0.90) but generally showed higher concentrations (up to 143% difference in North India).
- Regional Highlights: High BC concentrations were observed in Southeast Asia, South Asia, southern Africa, and South America. Absorbing (CAI) and scattering (CNAI) dust components were concentrated in the Sahara Desert, Middle East, and South Asia.
Limitations and Future Work
- MERRA-2 Underestimation: The study consistently found that MERRA-2 tends to underestimate BC and dust concentrations compared to the GRASP/AERONET retrievals, especially in regions with high aerosol loading. This highlights a need for further improvements in MERRA-2's aerosol modeling and assimilation.
- Regional Biases: Significant bias differences were noted in specific regions (e.g., sub-Sahel for BC, North India for dust), suggesting that MERRA-2's performance varies geographically.
- Assumed Refractive Indices: The GRASP/component approach relies on fixed complex refractive indices for aerosol components. While standard values are used, variations in the actual composition of these components could introduce uncertainties.
- Future Work: The authors emphasize the need for more effort to improve MERRA-2's component products in regions with large discrepancies. The newly created global dataset is expected to be invaluable for improving aerosol modeling and the assessment of radiative forcing. The authors also point out that the retrievals of absorbing components with very low aerosol loading can have extremely high uncertainty.
Summary updated