Near‐Cloud Aerosol Retrieval Using Machine Learning Techniques, and Implied Direct Radiative Effects. Issue 20 (18th October 2022)

• Record Type: Journal Article
• Authors: Yang, C. Kevin; Chiu, J. Christine; Marshak, Alexander; Feingold, Graham; Várnai, Tamás; Wen, Guoyong; Yamaguchi, Takanobu; Jan van Leeuwen, Peter
• Is Part Of: Geophysical research letters. Volume 49:Issue 20(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-10-18
• Subjects: aerosol remote sensing -- machine learning -- aerosol direct radiative effect -- transition zone; Geophysics -- Periodicals; Planets -- Periodicals; Lunar geology -- Periodicals; 550
• DOI: 10.1029/2022GL098274
• Languages: English
• ISSNs: 0094-8276

Abstract

Abstract: There is a lack of satellite‐based aerosol retrievals in the vicinity of low‐topped clouds, mainly because reflectance from aerosols is overwhelmed by three‐dimensional cloud radiative effects. To account for cloud radiative effects on reflectance observations, we develop a Convolutional Neural Network and retrieve aerosol optical depth (AOD) with 100–500 m horizontal resolution for all cloud‐free regions regardless of their distances to clouds. The retrieval uncertainty is 0.01 + 5%AOD, and the mean bias is approximately −2%. In an application to satellite observations, aerosol hygroscopic growth due to humidification near clouds enhances AOD by 100% in regions within 1 km of cloud edges. The humidification effect leads to an overall 55% increase in the clear‐sky aerosol direct radiative effect. Although this increase is based on a case study, it highlights the importance of aerosol retrievals in near‐cloud regions, and the need to incorporate the humidification effect in radiative forcing estimates. Plain Language Summary: The presence of aerosols can heat or cool the atmosphere, depending on their interactions with clouds and radiation. These interactions remain one of the primary sources of uncertainty in climate change predictions. To understand the role of aerosols in climate and their interactions with clouds, reflectance measurements from satellites have been used to retrieve aerosol properties. However, these retrievals are typically available in regions far from clouds, but not in the vicinity of clouds, because the observed reflectance is dominated by nearby cloud scattering rather than by aerosols. Since more than half of cloud‐free regions are within 4 km of low clouds, it is crucial to characterize the properties of aerosols near clouds. To tackle the issue, we developed a machine‐learning based retrieval method. The new method characterizes cloud radiative effects, removes them from the observed reflectance, and then retrieves aerosol properties even in the vicinity of clouds. The retrieval uncertainty is comparable to benchmark products. This newly added capability allows us to fill the critical gap in current aerosol observations and better quantify how aerosols influence the Earth's radiation budget. Key Points: A convolutional neural network is used to retrieve aerosol optical depth (AOD) with an uncertainty of 0.01 + 5%AOD in all cloud‐free regions Due to aerosol hygroscopic growth, the optical depth of aerosols near clouds can be enhanced by 100% compared to those far from clouds The enhancement in AOD near clouds leads to an overall 55% increase in clear‐sky aerosol direct radiative effects