Clouds Decoded: Understanding Mixed-Phase Clouds with High Resolution Multispectral Data

Clouds and their radiative effects remain one of the most pernicious unknowns in  our predictions of the climate. Climate models are particularly affected by uncertainties around mixed-phase clouds (MPCs) consisting of both super-cooled liquid droplets and ice crystals. Inaccurate measurements of the liquid water content can lead to an under or overestimation of the warming properties of these clouds. Moreover, we lack adequate constraints and parameterisations of the spatial heterogeneity of water/ice mixtures, and the distribution of ice crystal sizes within MPCs. 

Traditional cloud-monitoring satellites are able to retrieve physical cloud properties pertinent to these unknowns via, for example, LIDAR and radar sensors, but must necessarily treat MPCs as homogeneously mixed at scales below their resolution, which is on the order of 100s of metres to kilometres per pixel. Meanwhile, cloudy imagery from multispectral satellites with high spatial resolution, such as Sentinel-2, is treated as a waste product, with the ~60% of cloudy pixels left to gather dust in the archive. Multiple barriers exist that make these multispectral satellites difficult to use for ice cloud property retrievals, including their lack of thermal information, their tendency to mostly observe over land and not the ocean, their infrequent revisit times, and their sun-synchronous orbits which mean they only observe at 10 am local time. Nevertheless, these sensors offer a unique angle from which to study clouds, which can complement existing and future cloud-specialised sensors.

Here, we present early results of the Clouds Decoded project, funded by the Advanced Research + Invention Agency (ARIA), which seeks to help the community to tackle some of the key unknowns related to MPCs and ice clouds. Clouds Decoded aims to retrieve several physical cloud properties including cloud top height, optical depth (for optically thin clouds), ice/water ratio and ice particle effective radius, all at very fine resolution. This is being done at a massive scale, with the goal of processing several hundred terabytes of Sentinel-2 data from across the globe during the project. In this presentation, we will focus on a handful of case-studies which demonstrate how our data can be of use to the community. In particular, statistical relationships between cloud top temperature (inferred from height) and ice properties, alongside spatial frequency analyses, will be leveraged to describe the complex processes occurring in MPCs.