Clouds over the Southern Ocean play an important role in Earth’s climate, especially in how they reflect sunlight and contribute to rainfall. Many of these clouds form in organized patterns known as 'mesoscale cellular convection,' which appear as open or closed cell structures. These patterns affect cloud behavior, including their height, droplet size, and how much rain they produce.

In this study, we used a machine learning model to identify open and closed cell cloud types from satellite images over the ocean south of Australia. We compared different satellite and reanalysis datasets to see how well they capture differences between these cloud types. We found that some commonly used satellite products, especially those relying on passive sensors, often miss key differences, such as cloud height, between open and closed cells. In contrast, data from more advanced sensors like CALIPSO do show these differences.

We also looked at droplet size and number, and how rainfall is represented across datasets. To check these results, we compared them with one year of local observations from the CAPE-k campaign. These local measurements help confirm which satellite products are more reliable for studying these important cloud patterns.