 Satellite observations confirm aerosol-induced decreases in cloud droplet size in large areas compared to nearby less polluted areas  Aerosol-induced cloud water increases and decreases compensate each other at large spatial scales similarly to ship-track-like perturbations  On average, Twomey effect is more important than the effect of cloud water changes for the aerosol radiative forcing of the Earth's climate

The Copernicus Sentinel-2 mission operated by the European Space Agency (ESA) provides comprehensive and continuous multi-spectral observations of all the Earth’s land surface since mid-2015. Clouds and cloud shadows significantly decrease the usability of optical satellite data, especially in agricultural applications; therefore, an accurate and reliable cloud mask is mandatory for effective EO optical data exploitation. During the last few years, image segmentation techniques have developed rapidly with the exploitation of neural network capabilities. With this perspective, the KappaMask processor using U-Net architecture was developed with the ability to generate a classification mask over northern latitudes into the following classes: clear, cloud shadow, semi-transparent cloud (thin clouds), cloud and invalid. For training, a Sentinel-2 dataset covering the Northern European terrestrial area was labelled. KappaMask provides a 10 m classification mask for Sentinel-2 Level-2A (L2A) and Level-1C (L1C) products. The total dice coefficient on the test dataset, which was not seen by the model at any stage, was 80% for KappaMask L2A and 76% for KappaMask L1C for clear, cloud shadow, semi-transparent and cloud classes. A comparison with rule-based cloud mask methods was then performed on the same test dataset, where Sen2Cor reached 59% dice coefficient for clear, cloud shadow, semi-transparent and cloud classes, Fmask reached 61% for clear, cloud shadow and cloud classes and Maja reached 51% for clear and cloud classes. The closest machine learning open-source cloud classification mask, S2cloudless, had a 63% dice coefficient providing only cloud and clear classes, while KappaMask L2A, with a more complex classification schema, outperformed S2cloudless by 17%.

Aerosols, acting as cloud condensation nuclei, modulate the radiative properties of clouds. The influence of aerosols on Earth's climate is the most uncertain mechanism of anthropogenic climate forcing (Forster et al., 2021). Our limited knowledge about the extent to which aerosol offset the greenhouse gas warming impedes estimating the sensitivity of the Earth's climate to anthropogenic radiative forcing (Stevens et al., 2016), limiting the reliability of future climate projections.Most of the uncertainty associated with aerosol climate forcing comes from the poorly understood aerosol-cloud interactions (Bellouin et al., 2020). At fixed liquid water content, aerosols raise the cloud albedo by increasing cloud droplet number concentration (CDNC) and decreasing the size of cloud droplets, known as the first aerosol indirect effect (Twomey, 1974). Other aerosol indirect effects concern the changes in cloud thickness, cloud coverage, and lifetime (e.g., Albrecht, 1989). These adjustments can induce further changes in the albedo of a cloud field.Environmental conditions influence aerosol-cloud interactions. However, as meteorological conditions affect both aerosols and clouds (Liu et al., 2018), it can be difficult to disentangle the influence of anthropogenic aerosols on clouds (Quaas et al., 2020;Stevens & Feingold, 2009). The susceptibility of cloud droplet number concentration to aerosols depends on the geographical region and cloud regime (

Not only anthropogenic greenhouse gases but also microscopic anthropogenic air pollution particles called aerosols influence Earth's climate. Despite strong research efforts, the aerosol forcing of Earth's climate is still poorly quantified compared to the greenhouse gas (GHG) induced forcing (Bellouin et al., 2020). Considering multiple lines of evidence, total aerosol effective radiative forcing is estimated to be from −2.0 to −0.4 W/m 2 with a 90% likelihood (Bellouin et al., 2020). The fact that aerosols offset a poorly quantified fraction of GHG-induced positive radiative forcing makes it challenging to estimate the sensitivity of the Earth's climate to anthropogenic radiative forcing (Stevens et al., 2016) and improve the reliability of climate projections.Besides the direct radiative interactions, aerosols act as cloud condensation nuclei (CCN) and modulate cloud properties. The climate forcing caused by aerosol impacts on clouds is especially poorly quantified (Bellouin et al., 2020). The first indirect effect of aerosols, also called the Twomey effect (Twomey, 1974), refers to the increased cloud droplet number concentration (CDNC) in clouds. The Twomey effect raises the cloud albedo and induces a cooling effect on the Earth's climate. The magnitude of the physically well understood Twomey effect is relatively uncertain (Quaas et al., 2020), but the cooling effect is confirmed by multiple lines of evidence (Bellouin et al., 2020).The second aerosol indirect effect concerns the liquid water path (LWP) and cloud fraction response to increased CDNC (Albrecht, 1989;IPCC, 2013). LWP can increase due to suppressed collision-coalescence efficiency leading to suppressed precipitation (Albrecht, 1989) and decrease due to aerosol-enhanced entrainment (Ackerman et al., 2004;Bretherton et al., 2007;Wood, 2007). However, the net cloud water response to aerosols remains relatively poorly constrained (Bellouin et al., 2020).

<p>Aerosols offset poorly quantified fraction of anthropogenic greenhouse gas warming, whereas the aerosol impact on clouds is the most uncertain mechanism of anthropogenic climate forcing. In this research, we extend satellite observations of polluted cloud tracks from Toll et al. (2019, Nature, https://doi.org/10.1038/s41586-019-1423-9) with analysis of larger scale polluted cloud areas detected in MODerate-resolution Imaging Spectroradiometer satellite images. We demonstrate that large-scale anthropogenic aerosol-induced cloud perturbations exist at various major industrial aerosol source regions. The areal extent of the polluted cloud areas detected in MODIS satellite images extended to hundreds by hundreds of kilometres. Polluted clouds detected in satellite images in the global anthropogenic air pollution hot spot of Norilsk, Russia, and in other regions show close compensation between aerosol-induced cloud water increases and decreases. On average, there is relatively weak decrease in cloud water in the large areas with strong decreases in cloud droplet radii. This is in very good agreement with previous results based on small-scale polluted cloud tracks (Toll et al., 2019) and strongly disagrees with unidirectionally increased liquid water path in global climate models.</p>