A cloud’s droplet size—and its internal structure—from monochrome multi-angle imaging

Cloud droplet size plays a key role in precipitation formation, the cloud radiative effect, and the coupling of clouds to atmospheric turbulence. Due to this confluence of processes, measurements of droplet size are critical to our understanding of the climate system. Satellite remote sensing is the primary means of collecting these measurements over the globe and it is generally considered that multi-spectral or polarized measurements are required to isolate droplet size from droplet concentration. Here, we challenge this paradigm. We hypothesize that key information about the droplet size is encoded in the multi-scale structure of multi-angle imagery enabling the retrieval of droplet size from monochrome non-polarized measurements in the visible spectrum. We prove this hypothesis using forward and inverse (i.e., tomographic) 3D radiative transfer modeling based on realistic cloud models. We demonstrate that a vertical profile of droplet effective radius and the 3D volume extinction coefficient of the cloud can be retrieved using high-resolution (↭ 50 m) imagery at nine viewing angles using only the red channel. We attribute this success to the turbulent nature of cumuliform clouds that produces tenuous, optically thin regions at cloud edge that are only detectable at high spatial resolution. Observations of these regions are sensitive to scattering phase function, hence to droplet size, due to the dominance of low-order scattering. Finally, we verify that the addition of multi-spectral measurements in the shortwave infrared improves the tomographic retrieval of both the 1D droplet effective radius profile and the 3D extinction coefficient field through their improved sensitivity to regions where higher-order scattering dominates.