Che Huizheng¹， Zheng Yu¹， Zhao Hengheng¹， Zhang Xindan¹， Gui Ke¹， Li Lei¹， Miao Yucong¹， Zhao Hujia²， Shang Nanxuan¹，Peng Shiya¹， Zhu Jibiao¹， Zhang Xutao¹， Liu Shuhui¹， Zhang Haoling¹， Ju Mengyu¹，Yang Zhou¹， Ma Jingrui¹， Zhang Xiaoye¹

1.State Key Laboratory of Severe Weather Meteorological Science and Technology， Chinese Academy of Meteorological Sciences， Beijing 100081， China； 2.Shenyang Institute of Atmospheric Environment， China Meteorological Administration， Shenyang 110166， China

Abstract

Significance Atmospheric aerosols consist of suspended solid and liquid particles ranging from 0.001 μm to 100 μm in aerodynamic diameter， originating from both natural processes including dust emission， sea spray and volcanic eruption， and anthropogenic activities such as industrial exhaust， vehicle emissions， and fossil‑fuel combustion. With typical atmospheric lifetimes ranging from several days to one week， aerosols can undergo long‑range transport driven by atmospheric circulation， exerting profound impacts on near‑surface air quality， regional water cycles， and global climate systems. As emphasized in the IPCC Sixth Assessment Report， anthropogenic aerosols contribute substantial negative radiative forcing with considerable uncertainty， and aerosol–cloud interactions remain one of the dominant error sources in modern climate projections. Aerosols modify the Earth–atmosphere energy budget through direct radiative effects via scattering and absorbing solar and terrestrial radiation， as well as indirect radiative effects by acting as cloud condensation nuclei and ice nuclei to alter cloud microphysical properties and precipitation efficiency. Aerosol optical properties， represented by core parameters including aerosol optical depth （AOD）， Ångström exponent （AE）， single‑scattering albedo （SSA）， phase function， and asymmetry factor， serve as critical bridges linking microphysical‑chemical particle characteristics to environmental and climatic consequences. These optical parameters are predominantly acquired through ground‑based and satellite remote sensing observations. Although global observation networks have been established， aerosol properties exhibit strong spatial heterogeneity and seasonal variability across China. Therefore， a comprehensive understanding of aerosol optical properties is essential for reducing uncertainties in climate models， evaluating air‑pollution hazards， and supporting the implementation of carbon peaking and carbon neutrality strategies.

Progress This review systematically synthesizes recent advances in domestic research on ground‑based observations， satellite remote sensing， and artificial intelligence technologies. For ground‑based observations， China has constructed independent ground networks dominated by the China Aerosol Remote Sensing Network （CARSNET）， alongside CSHNET and SONET， with over 124 CARSNET stations covering urban agglomerations， dust source regions， and the Qinghai-Xizang Plateau. A hierarchical calibration system relying on the high‑altitude Waliguan baseline station has been developed， incorporating improved Langley calibration algorithms and integrating‑sphere sky‑radiation calibration to ensure AOD measurement accuracy better than 0.01. Passive sun‑photometer observations reveal nationwide spatiotemporal patterns of aerosol properties and meteorological modulation mechanisms， while ground‑based lidar enables continuous day‑and-night detection of vertical aerosol structures， supporting investigations of dust transport， haze‑fog evolution， and cross‑border aerosol import. Moreover， ground‑based optical measurements are combined with retrieval algorithms such as GRASP‑Component to constrain aerosol component fractions and quantify direct aerosol radiative forcing. In satellite remote‑sensing research， international satellite products （MODIS， MISR， CALIPSO， etc.） have been validated and applied across diverse land surfaces in China. Domestic satellite payloads from the Gaofen， Fengyun， and environmental‑monitoring satellite series provide multi‑angle polarimetric and geostationary observations. Chinese researchers have optimized dark‑target， deep‑blue and GRASP‑based retrieval algorithms， and evaluated the regional performance of MERRA‑2 and CAMS aerosol reanalysis datasets. Multi‑source fusion approaches combining ground measurements， satellite retrievals， and reanalysis data， particularly machine‑learning‑based fusion frameworks， mitigate data gaps induced by cloud contamination. In terms of artificial intelligence applications， data‑driven methods alleviate the limitations of traditional physical inversion approaches constrained by prior assumptions. Advanced models enable intelligent retrieval of aerosol parameters，seamless spatiotemporal reconstruction of key atmospheric variables such as PM₂.₅ and visibility，and hybrid physical‑data‑driven aerosol forecasting. Deep‑learning surrogate models also accelerate time‑consuming radiation transfer and aerosol chemistry calculations and correct simulation biases within numerical chemical transport models.

Conclusions and Prospects Despite these remarkable achievements， multiple bottlenecks remain. Ground‑based systems still face limitations， including insufficient night‑time observation capacity， uneven spatial distribution of monitoring sites especially over remote complex underlying surfaces， and incomplete parameterization for aerosol-radiation-boundary‑layer coupling processes. Domestic satellite systems are still limited in retrieving high-resolution aerosol microphysical and chemical parameters. Mature multi‑satellite collaborative inversion frameworks have yet to be established， and large retrieval errors remain over bright and complicated land surfaces. For artificial intelligence methodologies， poor model interpretability restricts operational deployment， and interdisciplinary barriers hinder the combination of physical aerosol knowledge and data‑driven algorithms. Looking ahead， future priorities cover four dimensions. First， observation infrastructures should be upgraded to strengthen integrated diurnal monitoring and collaborative multi‑platform observations. Second， inversion and modelling algorithms should be optimized to improve retrievals of fine aerosol parameters and refine aerosol-cloud-radiation parameterization schemes. Third， the operational transformation of artificial intelligence technologies should be promoted by developing physically informed interpretable models and overcoming disciplinary boundaries. Fourth， frontier scientific studies focusing on carbonaceous and dust aerosol evolution under the dual‑carbon background should be advanced to disentangle linkage mechanisms among anthropogenic emission reduction， aerosol variations， and climate feedbacks， thereby providing scientific support for environmental governance and climate change assessment.

Keywords： optical property of atmospheric aerosol； aerosol optical depth； ground-based observation； satellite remote sensing； artificial intelligence

Get Citation: Che Huizheng， Zheng Yu， Zhao Hengheng， et al. Research progress on optical properties of atmospheric aerosols： from conventional observations to intelligent remote sensing technologies ［J］. China Powder Science and Technology， 2027， 33（2）： 108-126.

Received：2026-07-15， Revised： 2026-09-26， Online： 2026-10-11。

Funding： The research was supported by the National Key R&D Program of China (Grant No. 2024YFB3908701), the National Natural Science Foundation of China (Grant Nos. 42375188, U2542218, 42322504, 42575203, and 42475193), the Science and Technology Development Fund of Chinese Academy of Meteorological Sciences (Grant No. 2025KJ001), and the Basic Research Fund of Chinese Academy of Meteorological Sciences (Grant No. 2024Z006).

DOI：10.13732/j.issn.1008-5548.2027.02.009

CLC No.: P412.27；TB4

Type Code：A

Serial No.:1008-5548（2027）02-0108-19