Object-based characterization of Tropical Trade Cumuli During the EUREC4A/ATOMIC Field Campaign using Sentinel-2 observations

In January and February 2020, the joint EUREC⁴A/ATOMIC field campaign took place in the Tropical Atlantic near Barbados, with the goal to advance our understanding of the interplay between clouds, convection and circulation including their role in climate change. Within the scope of this campaign, several unique satellite-based datasets have been collected, including very high-spatial resolution multispectral images with 10x10m² pixel size acquired by polar-orbiting Copernicus Sentinel-2 satellites. In this presentation, the first analysis of these high-resolution observations focused on tropical trade cumuli is given. This cloud type is characterized by small-scale spatiotemporal variability that is unresolved at the spatial resolution of current meteorological satellite imagers.

Using the high-resolution Sentinel-2 observations, we will show the clouds can be considered as individual objects with associated properties, such as shape and size parameters, but also their mean radiative properties. A novel technique will be presented for deriving cloud height from Sentinel-2 observations, which exploits the geometric relationship between cloud objects and their shadows. Furthermore, the cloud fraction and cloud size distribution are calculated for various trade cumulus scenes. The uncertainties arising from choices in our cloud detection scheme will be discussed.

We show that a substantial fraction of clouds has equivalent diameters below the pixel size of commonly used meteorological satellite instruments. Consequently, and consistent with previous studies, the cloud size distribution and domain-average cloud fraction from coarse-resolution satellite imagers are shown to be biased and highly sensitive to pixel resolution. In addition, a large fraction of pixels identified as cloudy contains significant clear-sky contributions, and it is no longer possible to characterize clouds as objects. We will discuss how this affects the accuracy of cloud property determination and biases estimates of cloud radiative forcing.