ML-AOMP: Global Aerosol Optical and Microphysical Properties from PACE-SPEXone Satellite

1.7 GB of aerosol property data retrieved using a machine learning framework from PACE-SPEXone satellite observations. The ML-AOMP framework, developed by Zigeng Song, achieved a coefficient of determination (R²) exceeding 0.8 and errors under 10% compared to operational algorithms. It provides global retrievals of spectral aerosol optical depth, single scattering albedo, refractive index, and mode-resolved microphysical parameters.

Use Cases

• Validating aerosol retrieval algorithms based on spectral aerosol optical depth (AOD) and single scattering albedo (SSA) data.
• Analyzing global spatial patterns of aerosol properties to identify major source regions.
• Studying the relationship between satellite observables and microphysical parameters like effective radius and volume concentration.
• Assessing the performance of machine learning frameworks for geophysical parameter retrieval from polarimetric satellite data.

Strengths

• Demonstrated high accuracy with R² exceeding 0.8 and RMSE/MAPE under 10% compared to the operational RemoTAP algorithm.
• Includes a wide range of aerosol properties: spectral AOD, SSA, refractive index, and mode-resolved parameters for coarse, fine, and total modes.
• Validated against independent global networks (AERONET and SKYNET), suggesting robustness and generalizability.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.
• Data is stored in HDF5 format (H5), which may require specific libraries for access and processing.

Provenance

Source

PACE-SPEXone satellite observations, with training/validation from AERONET and SKYNET ground-based networks.

Collection Method

Retrieved using the ML-AOMP machine learning framework trained on global AERONET measurements.

Freshness

Last updated 2026-04-30 11:46:39; freshness should be verified.

Geography

Global coverage, as indicated by analysis of global spatial patterns and major aerosol source regions.