Computed cloud tomography (CCT) is a promising new approach in remote sensing of large vertically-developed clouds using space-based imaging sensors with pixel scales in the 100s of meters. Prime examples are the Multi-angle Imaging Spectro-Radiometer (MISR) and the MODerate-resolution Imaging Spectrometer (MODIS), both on NASA’s flagship Terra satellite. Current operational cloud property retrievals in the solar spectrum, which are grounded in 1D radiative transfer (RT) predictions for reflected radiation emanating from a single pixel and view angle at just two wavelengths. In sharp contrast, CCT embraces the 3D nature of real clouds and exploits collocated multi-spectral/multi-angle/multi-pixel data to recover volumetric information. However, CCT has been demonstrated so far only for rather small clouds using airborne sensors with ~20 m pixels.
A first step in the right direction was recently taken by Forster et al. (2021) who defined the “veiled core” (VC) of large opaque clouds as the optically deep region where detailed 3D structure of the cloud has little impact on the multi-angle/multi-spectral images as long as the mean VC extinction coefficient and any significant cloud-scale gradient are preserved. Quantitatively, the difference between radiance fields escaping the clouds is commensurate with sensor noise when said clouds differ only in the small-scale distribution of extinction inside their VCs.
An important corollary for the large and ill-posed CCT inverse problem is that the only unknowns of interest for the whole VC are its mean extinction coefficient and any potential cloud-scale vertical trend in that property. Another ramification for CCT algorithms currently under development for space-based sensor data is that the forward 3D RT model driving the inversion may be vastly simplified in the VC to gain efficiency. We explore here that possibility, assuming radiative diffusion as the simplified RT for the VC. On the way, we describe the relevant RT physics that unfold in the VC and in the outer shell (OS) of the cloud, where detailed spatial structure does matter for image formation. This includes control by the VC of the cloud-scale contrast between the brightness of illuminated and shaded cloud sides, as well as the gradual blurring of spatial structure via directional diffusion with increasing optical distance into the OS.
Transport space is the merger of 3D physical space and 2D direction space. Cloud image formation involves radiative diffusion processes (i.e., random walks) in each of these subspaces, depending on what transport regime prevails. Fortunately for the future of CCT, and of passive cloud remote sensing in general, there is a clear spatial separation: asymptotic limit of radiative diffusion in the VC, standard RT in the OS. A hybrid forward model for CCT will make use of this fact of life in cloud image formation.
Reference:
Forster, L., et al. (2021). J. Atmos. Sci., 78, 155-166. doi:10.1175/JAS-D-19-0262.1