Aerosol optical depth (AOD) is an important measure of atmospheric aerosols, with applications in air quality monitoring and climate research. However, existing retrieval methods face challenges under complex conditions and often make limited use of satellite spectral and polarization information.

In a study published in IEEE Transactions on Geoscience and Remote Sensing, researchers from the Hefei Institutes of Physical Science of the Chinese Academy of Sciences developed an intelligent algorithm to improve AOD retrieval from China's spaceborne polarimetric remote sensing data.

Researchers developed an attention-enhanced Kolmogorov–Arnold network (AKAN) to improve AOD retrieval from satellite observations. The network combines the learning ability of KANs with an attention mechanism, helping it focus on important information from multispectral polarimetric data.

Researchers tested AKAN using observations from Particulate Observing Scanning Polarimeter (POSP) onboard Chinese Gaofen-5B satellite. By combining POSP observations with ground-based measurements from global Aerosol Robotic Network sites and supplemental regional sun-photometer observations, they constructed a dataset of 243,584 matched samples.

AKAN achieved accurate global AOD retrieval, with an R² value of 0.9336 and most results meeting the expected requirements.

Using SHapley Additive exPlanations method, researchers examined how the model make its predictions. They showed that important spectral bands and scattering-angle information selected by the model agreed well with known atmospheric processes, indicating that the artificial intelligence model provides accurate results and useful information for understanding aerosol observations.

Further tests showed that the model maintained good accuracy and adaptability under various atmospheric and surface conditions.

This study offers a promising approach for improving aerosol monitoring with satellite remote sensing data.